In vitro characterization of the enzymes involved in TDP-D-forosamine biosynthesis in the spinosyn pathway of Saccharopolyspora spinosa

Article abstract of DOI:10.1021/ja0771383

Forosamine (4-dimethylamino)-2,3,4,6-tetradeoxy-β-d-threo- hexopyranose) is a highly deoxygenated sugar component of several important natural products, including the potent yet environmentally benign insecticide spinosyns. To study D-forosamine biosynthesis, the five genes (spnO, N, Q, R, and S) from the spinosyn gene cluster thought to be involved in the conversion of TDP-4-keto-6-deoxy-D-glucose to TDP-D-forosamine were cloned and heterologously expressed, and the corresponding proteins were purified and their activities examined in vitro. Previous work demonstrated that SpnQ functions as a pyridoxamine 5′-monophosphate (PMP)-dependent 3-dehydrase which, in the presence of the cellular reductase pairs ferredoxin/ferredoxin reductase or flavodoxin/flavodoxin reductase, catalyzes C-3 deoxygenation of TDP-4-keto-2,6-dideoxy-D-glucose. It was also established that SpnR functions as a transaminase which converts the SpnQ product, TDP-4-keto-2,3,6-trideoxy-D- glucose, to TDP-4-amino-2,3,4,6-tetradeoxy-D-glucose. The results presented here provide a full account of the characterization of SpnR and SpnQ and reveal that SpnO and SpnN functions as a 2,3-dehydrase and a 3-ketoreductase, respectively. These two enzymes act sequentially to catalyze C-2 deoxygenation of TDP-4-keto-6-deoxy-D-glucose to form the SpnQ substrate, TDP-4-keto-2,6-dideoxy- D-glucose. Evidence has also been obtained to show that SpnS functions as the 4-dimethyltransferase that converts the SpnR product to TDP-D-forosamine. Thus, the biochemical functions of the five enzymes involved in TDP-D-forosamine formation have now been fully elucidated. The steady-state kinetic parameters for the SpnQ-catalyzed reaction have been determined, and the substrate specificities of SpnQ and SpnR have been explored. The implications of this work for natural product glycodiversification and comparative mechanistic analysis of SpnQ and related NDP-sugar 3-dehydrases E₁ and ColD are discussed.

Full text of DOI:10.1021/ja0771383

Published on Web 03/18/2008
In Vitro Characterization of the Enzymes Involved in
TDP-D-Forosamine Biosynthesis in the Spinosyn Pathway of
Saccharopolyspora spinosa
Lin Hong,† Zongbao Zhao,# Charles E. Melanc¸on III,‡ Hua Zhang,§ and
Hung-wen Liu*
DiVision of Medicinal Chemistry, College of Pharmacy, and Department of Chemistry and
Biochemistry, UniVersity of Texas at Austin, Austin, Texas 78712
Received September 14, 2007; E-mail: h.w.liu@mail.utexas.edu
Abstract: Forosamine (4-dimethylamino)-2,3,4,6-tetradeoxy-â-d-threo-hexopyranose) is a highly deoxy-
genated sugar component of several important natural products, including the potent yet environmentally
benign insecticide spinosyns. To study D-forosamine biosynthesis, the five genes (spnO, N, Q, R, and S)
from the spinosyn gene cluster thought to be involved in the conversion of TDP-4-keto-6-deoxy-D-glucose
to TDP-D-forosamine were cloned and heterologously expressed, and the corresponding proteins were
purified and their activities examined in vitro. Previous work demonstrated that SpnQ functions as a
pyridoxamine 5′-monophosphate (PMP)-dependent 3-dehydrase which, in the presence of the cellular
reductase pairs ferredoxin/ferredoxin reductase or flavodoxin/flavodoxin reductase, catalyzes C-3 deoxy-
genation of TDP-4-keto-2,6-dideoxy-D-glucose. It was also established that SpnR functions as a transami-
nase which converts the SpnQ product, TDP-4-keto-2,3,6-trideoxy-D-glucose, to TDP-4-amino-2,3,4,6-
tetradeoxy-D-glucose. The results presented here provide a full account of the characterization of SpnR
and SpnQ and reveal that SpnO and SpnN functions as a 2,3-dehydrase and a 3-ketoreductase, respectively.
These two enzymes act sequentially to catalyze C-2 deoxygenation of TDP-4-keto-6-deoxy-D-glucose to
form the SpnQ substrate, TDP-4-keto-2,6-dideoxy-D-glucose. Evidence has also been obtained to show
that SpnS functions as the 4-dimethyltransferase that converts the SpnR product to TDP-D-forosamine.
Thus, the biochemical functions of the five enzymes involved in TDP-D-forosamine formation have now
been fully elucidated. The steady-state kinetic parameters for the SpnQ-catalyzed reaction have been
determined, and the substrate specificities of SpnQ and SpnR have been explored. The implications of
this work for natural product glycodiversification and comparative mechanistic analysis of SpnQ and related
NDP-sugar 3-dehydrases E1 and ColD are discussed.
Introduction
enzymatic glycodiversification strategies for generating diversity
in natural product sugar structures5-9 has underscored the
Deoxysugars are commonly found as components of bioactive
natural products such as antimicrobial agents and anticancer
drugs1 and are often crucial for the activities of these com-
pounds.2,3 Because of the variety of deoxysugar structures found
in nature and their importance for natural product bioactivity,
much effort has been devoted to the elucidation of their
biosynthetic pathways.4-7 The proven feasibility of genetic and
importance of collecting and characterizing deoxysugar bio-
synthetic components for use in these endeavors. D-Forosamine
(4-dimethylamino)-2,3,4,6-tetradeoxy-â-D-threo-hexopyra-
nose, 1, Scheme 1) is a highly deoxygenated sugar found in
the structures of many bioactive natural products, including the
insecticidal spinosyns A and D (2, 3)10 and butenylspinosyn
(4),11 the macrolide antibiotic spiramycin (5),12 the naphtho-
quinone antibiotic forosaminylgriseucin A (6),13 and the im-
munosuppressive agent dunaimycin (7).14
† Current Address: GNF, Inc., La Jolla, CA 92037.
# Current address: Dalian Institute of Chemical Physics, CAS, Dalian
116023, P. R. China.
‡ Current address: Department of Chemistry, Scripps Research Institute,
La Jolla, CA 92037.
(8) Blanchard, S.; Thorson, J. S. Curr. Opin. Chem. Biol. 2006, 10, 263-271.
(9) Thibodeaux, C. J.; Liu, H.-w. Pure Appl. Chem. 2007, 79, 785-799.
(10) Waldron, C.; Matsushima, P.; Rosteck, P. R., Jr.; Broughton, M. C.; Turner,
J.; Madduri, K.; Crawford, K. P.; Merlo, D. J.; Baltz, R. H. Chem. Biol.
2001, 8, 487-499.
(11) Hahn, D. R.; Gustafson, G.; Waldron, C.; Bullard, B.; Jackson, J. D.;
Mitchell, J. J. Ind. Microbiol. Biotechnol. 2006, 33, 94-104.
(12) Blondelet-Rouault, M.-H.; Dominguez, H.; Darbon-Rongere, E.; Gerbaud,
C.; Gondran, A.; Karray, F.; Lacroix, P.; Oestreicher-Mermet, B. N.;
Pernodet, J.-L.; Tuphile, K. PCT Int. Appl. WO 2004033689, 2004.
(13) Maruyama, M.; Nishida, C.; Takahashi, Y.; Naganawa, H.; Hamada, M.;
Takeuchi, T. J. Antibiot. 1994, 47, 952-954.
(14) Hochlowski, J. E.; Mullally, M. M.; Brill, G. M.; Whittern, D. N.; Buko,
A. M.; Hill, P.; McAlpine, J. B. J. Antibiot. 1991, 44, 1318-1330.
§ Current address: Institute of Biosciences and Technology, Texas A&M
University Health Science Center, Houston, TX.
(1) Johnson, D. A.; Liu, H.-w. ComprehensiVe Natural Products Chemistry;
Elsevier: Amsterdam, New York, 1999; Vol. 3, pp 311-365.
(2) Kren, V.; Martinkova, L. Curr. Med. Chem. 2001, 8, 1303-1328.
(3) Weymouth-Wilson, A. C. Nat. Prod. Rep. 1997, 14, 99-110.
(4) Trefzer, A.; Salas, J. A.; Bechthold, A. Nat. Prod. Rep. 1999, 16, 283-
299.
(5) He, X.; Liu, H.-w. Annu. ReV. Biochem. 2002, 71, 701-754.
(6) Salas, J. A.; Mendez, C. J. Mol. Microbiol. Biotechnol. 2005, 9, 77-85.
(7) Thibodeaux, C.; Melanc¸on, C. E., III; Liu, H.-w. Nature 2007, 446, 1008-
1016.
9
4954
J. AM. CHEM. SOC. 2008, 130, 4954-4967
10.1021/ja0771383 CCC: $40.75 © 2008 American Chemical Society

TDP-D-Forosamine Biosynthesis in the Spinosyn Pathway
A R T I C L E S
Scheme 1. Structures of Known Natural Products Containing D-forosamine Moieties
Spinosyns A and D (2, 3, Scheme 1) are the major natural
products made by the actinomycete Saccharopolyspora spinosa
and are commercially marketed as the insecticide Spinosad (also
known as Tracer or Naturalyte). These compounds are composed
of an unusual tetracyclic macrolide and two deoxysugars,
D-forosamine (1) and 2,3,4-tri-O-methyl-L-rhamnose. More than
20 minor spinosyn congeners that vary with respect to the
aliphatic substituents at C-6, C-16, and C-21, and by the degree
of methylation of the 2-, 3-, and 4-hydroxyl groups of the
L-rhamnose moiety and of the 4-amino group of D-forosamine,
are also produced by S. spinosa.15 Structure-activity studies
have shown that both D-forosamine and 2,3,4-tri-O-methyl-L-
rhamnose moieties of spinosyns are critical for their insecticidal
activities.16,17
Sequencing of the spinosyn biosynthetic gene cluster in S.
spinosa (Figure 1) and subsequent sequence analysis combined
with the results of gene disruption studies led to the proposal
that five genes, spnO, spnN, spnQ, spnR, and spnS, are involved
in the conversion of TDP-4-keto-6-deoxy-D-glucose (10) to
TDP-D-forosamine (16). The spnO gene was predicted to encode
a TDP-sugar 2,3-dehydratase and spnN to encode a TDP-sugar
3-ketoreductase that convert 10 to 11 and 11 to 12, respectively.
The spnQ gene was predicted to encode a TDP-sugar 3-dehy-
drase that converts 12 to 13, while spnR and spnS were assigned
as TDP-sugar aminotransferase and TDP-sugar methyltrans-
ferase converting 13 to 14 and 14 to 16, respectively. The spnP
product was proposed to transfer D-forosamine from 16 to the
hydroxyl group at C-17 of the tetracyclic pseudoaglycones (17/
18) to give 2/3 (Scheme 2).10 Compound 10 is formed by
thymidylylation of glucose-1-phosphate (8) to form TDP-D-
glucose (9) followed by 4,6-dehydration of 9. Normally, glucose-
1-phosphate thymidylyltransferase (Ep) and TDP-D-glucose-4,6-
dehydrase (Eod), which catalyze these reactions, are encoded
by genes that are present in the glycosylated natural product
gene clusters. However, the corresponding genes are absent in
the spinosyn gene cluster. Instead, genes encoding these
activities, designated gtt (glucose-1-phosphate thymidyltrans-
ferase) and gdh (TDP-D-glucose-4,6-dehydrase), respectively,
are found elsewhere in the genome. Gtt and Gdh have been
shown by gene disruption experiments to supply 10 to both
spinosyn biosynthesis and cell wall polysaccharide formation
in S. spinosa.18
Among the five assigned forosamine genes, the translated
spnO sequence is similar to several sugar 2,3-dehydratases,
including TylX3 (44% identity), EvaA (45% identity), Gra27
(42% identity), and OleV (48% identity), while that of spnN
shares significant sequence identity with 3-ketoreductases Gra26
(49% identity) and OleW (46% identity). These known 2,3-
dehydratases catalyze the conversion of 10 to an unstable
intermediate TDP-3,4-diketo-2,6-dideoxy-D-glucose (11), which
is then converted to TDP-4-keto-2,6-dideoxy-D-glucose (12) by
a 3-ketoreductase, such as Gra26 or OleW in the TDP-L-
rhodinose (19) and TDP-L-olivose (20) biosynthetic pathways,
respectively (Scheme 3A).19-21 The 3-ketoreductase TylC1,
which converts the TylX3 product (11) to TDP-4-keto-2,6-
(15) Sparks, T. C.; Kirst, H. A.; Mynderse, J. S.; Thompson, G. D.; Turner, J.
R.; Jantz, O. K.; Hertlein, M. B.; Larson, L. L.; Baker, P. J. et al. Proc.
Beltwide Cotton Conf. 1996, 692-696.
(16) Crouse, G.; Sparks, T. C. ReV. Toxicol. 1998, 2, 133-146.
(17) Sparks, T. C.; Thompson, G. D.; Kirst, H. A.; Hertlein, M. B.; Larson, L.
L.; Worden, T. V.; Thibault, S. T. J. Econ. Entomol. 1998, 91, 1277-
1283.
(18) Madduri, K.; Waldron, C.; Merlo, D. J. J. Bacteriol. 2001, 183, 5632-
5638.
(19) Chen, H.; Agnihotri, G.; Guo, Z.; Que, N. L. S.; Chen, X.; and Liu, H.-w.
J. Am. Chem. Soc. 1999, 121, 8124-8125.
(20) Chen, H.; Thomas, M. G.; Hubbard, B. K.; Losey, H. C.; Walsh, C. T.;
Burkart, M. G. Proc. Nat. Acad. Sci. 2000, 97, 11942-11947.
(21) Draeger, G.; Park, S.-H.; Floss, H. G. J. Am. Chem. Soc. 1999, 121, 2611-
2612.
Figure 1. The spinosyn biosynthetic gene cluster of Saccharopolyspora
spinosa showing the location and arrangement of genes encoding forosamine
biosynthesis.
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4955

A R T I C L E S
Hong et al.
Scheme 2. The Spinosyn Biosynthetic Pathway Highlighting the Biosynthesis of TDP-D-forosamine (16)a
a Note that compounds 22, 23, and 28 are not intermediates in the native biosynthetic pathway. Gtt is the TTP-dependent glucose-1-phosphate
Thymidylyltransferase converting 8 to 9, Gdh is the Tdp- D-glucose 4,6-dehydrase converting 9 to 10, SpnO is the TDP-4-keto-6-deoxy-D-glucose 2,3-
dehydratase converting 10 to 11, SpnN is the NAD(P)H-dependent TDP-3,4-diketo-2,6-dideoxy-D-glucose 3-ketoreductase converting 11 to 12, SpnQ is the
PMP-dependent TDP-4-keto-2,6-dideoxy-D-glucose 3-dehydrase converting 12 to 13, SpnR is the PLP, L-glutamate-dependent TDP-4-keto-2,3,6-trideoxy-
D-glucose 4-aminotransferase converting 13 to 14, and SpnS is the SAM-dependent TDP-4-amino-2,3,6-trideoxy-D-glucose N,N-dimethyltransferase sequentially
converting 14 to 15 and 15 to 16.
Scheme 3. (A) Structures of TDP-l-rhodinose (19), TDP-L-olivose (20), and TDP-4-keto-2,6-dideoxy-D-allose (21). (B) Transamination and
N-Methylation of the Isostere (24) of 14 by SpnR and SpnS, Respectively
dideoxy-D-allose (21, Scheme 3A), the C-3 epimer of 12, is
only distantly related to SpnN (14% identity). On the basis of
this information, it was predicted that SpnO and SpnN catalyze
the conversion of 10 to 11 and 11 to 12, respectively. Under in
Vitro conditions when a 3-ketoreductase is not present to convert
11 to a more stable product, 11 is known to rapidly degrade to
maltol (22) and TDP.19,21
ColD23,24 from the colitose pathway (49 and 26% identity to
SpnQ, respectively), both from Yersinia pseudotuberculosis.
SpnQ also exhibits high sequence similarity to a number of
putative NDP-sugar 3-dehydrases encoded in gene clusters of
2,3,6-trideoxyhexose-containing natural products, such as Gra23
from the granaticin pathway (68% identity),25 LanQ from the
landomycin pathway (67% identity),26 and UrdQ from the
SpnQ is homologous to several vitamin B6-dependent en-
zymes which catalyze C-3 deoxygenation of NDP-sugar sub-
(22) He, X.; Liu, H.-w. Curr. Opin. Chem. Biol. 2002, 6, 590-597.
(23) Alam, J.; Beyer, N.; Liu, H.-w. Biochemistry 2004, 43, 16450-16460.
(24) Cook, P. D.; Thoden, J. B.; Holden, H. M. Protein Sci. 2006, 15, 2093-
2106.
(25) Ichinose, K.; Bedford, D. J.; Tornus, D.; Bechthold, A.; Bibb, M. J.; Revill,
W. P.; Floss, H. G.; Hopwood, D. A. Chem. Biol. 1998, 5, 647-659.
strates, including the well-characterized CDP-4-keto-6-deoxy-
5,22
D-glucose 3-dehydrase E1
from the ascarylose biosynthetic
pathway and the GDP-4-keto-6-deoxy-D-mannose 3-dehydrase
9
4956 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

TDP-D-Forosamine Biosynthesis in the Spinosyn Pathway
A R T I C L E S
urdamycin pathway (68% identity),27 all of which are believed
to be involved in the formation of TDP-L-rhodinose (19). SpnQ
was therefore predicted to be a 3-dehydrase which converts 12
to TDP-4-keto-2,3,6-trideoxy-D-glucose (13).
glycodiversification using components of the forosamine bio-
synthetic machinery, and for comparative mechanistic and
structural studies of SpnQ and its 3-dehydrase homologues E1
and ColD.24
35
SpnR is homologous to numerous NDP-sugar aminotrans-
ferases, including DesI (26% identity) and TylB (27% identity)
whose functions have been established in Vitro.28,29 SpnR was
therefore predicted to be the 4-aminotransferase which converts
the SpnQ product 13 to TDP-4-amino-2,3,4,6-tetradeoxy-D-
glucose (14). Originally, we also considered the possibility that
the SpnR-catalyzed C-4 transamination might precede C-3
deoxygenation by SpnQ in the forosamine pathway. Assays of
SpnR in the forward direction using both the SpnN product 12
and in the reverse direction using a more stable phosphonate
mimic (24, Scheme 3B) of the proposed SpnR product (14)
showed that SpnR was in fact capable of catalyzing C-4
transamination of both substrates. However, SpnR is much more
efficient at transamination of 24 than 12 (more than 15-fold
difference in rate), strongly suggesting that SpnQ acts first in
the pathway.30 The fact that SpnQ efficiently converts the SpnN
product 12 to 13 further supports this idea.31 We have therefore
assigned SpnR the role of converting 13 to 14 in forosamine
biosynthesis. SpnS is homologous to a number of NDP-
aminosugar N,N-dimethyltransferases, including DesVI (46%
identity) and TylM1 (40% identity) from the desosamine and
mycaminose pathways, respectively, whose functions have been
demonstrated in Vitro.32-34 SpnS is therefore predicted to
catalyze the final step in forosamine biosynthesis, converting
the SpnR product 14 to TDP-D-forosamine (16).
Experimental Section
Materials. The spnO, N, Q, R, and S genes were amplified by
polymerase chain reaction (PCR) from the genomic DNA isolated from
Saccharopolyspora spinosa NRRL 18537, which was obtained from
the Agricultural Research Service Culture Collection (Peoria, IL).
Escherichia coli strain DH5R was purchased from Bethesda Research
Laboratories (Gaithersburg, MD). Vector pET24b(+) and pET28b(+)
and overexpression host E. coli BL21(DE3) were acquired from
Novagen (Madison, WI). Enzymes used for molecular cloning experi-
ments were products of Invitrogen (Carlsbad, CA). Ni-NTA agarose
and kits for DNA gel extraction and spin miniprep were obtained from
Qiagen (Valencia, CA). Pfu DNA polymerase and QuikChange Site-
Directed Mutagenesis Kit were purchased from Stratagene (La Jolla,
CA). The growth media components were acquired from Becton
Dickinson (Sparks, MD). Ferredoxin and ferredoxin reductase were
purchased from Sigma-Aldrich (St. Louis, MO). Unless otherwise
specified, all chemicals, reagents, and enzymes were purchased from
Sigma-Aldrich or Fisher Scientific (Pittsburgh, PA) and were used
without further purification. Bio-gel P2 resin, Econo-Pac 10 DG
desalting columns, and all reagents for sodium dodecyl sulfate-
polyacrylamide gel electrophoresis (SDS-PAGE) were acquired from
Bio-Rad Laboratories (Hercules, CA). Amicon YM-10 filtration
products were purchased from Millipore (Billerica, MA). The CarboPac
PA1 HPLC column was obtained from Dionex (Sunnyvale, CA). The
Superdex 200 HR 10/30 FPLC column was the product of Pharmacia
(Uppsala, Sweden). The SAX Adsorbosphere HPLC column was
purchased from Alltech (Deerfield, IL). Oligonucleotide primers for
cloning of all genes were prepared by Integrated DNA Technologies
(Coralville, IA).
General. Isolation of S. spinosa genomic DNA was performed
according to published procedures.36 Protein concentrations were
determined according to Bradford37 using bovine serum albumin as
the standard. The relative molecular mass and purity of enzyme samples
were estimated using SDS-PAGE as described by Laemmli.38 All
reagents and solvents were purchased from commercial sources and
were used without further purification unless otherwise noted. The NMR
spectra were acquired on a Varian Unity 300, 400, or 500 MHz
spectrometer, and chemical shifts are given relative to those for t-BuOH
(the chemical shifts are 1.27 and 31.2 ppm for 1H and 13C NMR,
respectively), and aqueous 85% H3PO4 (external, for 31P NMR), with
coupling constants reported in hertz (Hz). Flash chromatography was
performed on Lagand Chemical silica gel (230-400 mesh) by elution
with the specified solvents. Analytical thin-layer chromatography (TLC)
was carried out on Polygram Sil G/UV254 plates (0.25 mm). TLC spots
were visualized by heating plates previously stained with a solution of
phosphomolybdic acid (3% in EtOH). DNA sequencing and N-terminal
sequencing of purified enzymes were performed by the Core Facilities
of the Institute of Cellular and Molecular Biology at the University of
Texas at Austin. Chemical ionization (CI), fast-atom bombardment
(FAB), and electrospray ionization (ESI) mass spectra were recorded
by the MS facility in the Department of Chemistry and Biochemistry
of the University of Texas at Austin. The general methods and protocols
for recombinant DNA manipulations followed those described by
Sambrook et al.39 Kinetic data were analyzed by nonlinear fit using
Grafit5 (Erithacus Software Ltd.).
Herein, we report the overexpression, purification, and
characterization of SpnO, SpnN, and SpnS as well as additional
characterization of SpnQ and SpnR. We demonstrate via in Vitro
assays and NMR analysis of the purified SpnN product TDP-
4-keto-2,6-dideoxy-D-glucose (12) that SpnO and SpnN are the
2,3-dehydratase and 3-ketoreductase, which together catalyze
the conversion of 10 to 12. The 4-N-methyl transfer activity of
SpnS was also demonstrated by an in Vitro assay using the
chemically synthesized phosphonate mimic 24, NMR charac-
terization of the SpnS monomethylated product (25), and mass
spectrometric characterization of the phosphonate mimic of
TDP-D-forosamine (26) (Scheme 3B). We also assayed SpnR
using its physiological substrate 13 and determined the steady-
state kinetic parameters of the SpnQ-catalyzed reaction with
its authentic substrate 12. With these results, the functions of
all five enzymes in the TDP-D-forosamine biosynthetic pathway
have been established in Vitro, and the catalytic properties of
the enzymes have been explored. The information gained from
these studies lays the foundation for in Vitro and in ViVo
(26) Westrich, L.; Domann, S.; Faust, B.; Bedford, D.; Hopwood, D. A.;
Bechthold, A. FEMS Microbiol. Lett. 1999, 170, 381-387.
(27) Hoffmeister, D.; Ichinose, K.; Domann, S.; Faust, B.; Trefzer, A.; Drager,
G.; Kirschning, A.; Fischer, C.; Kunzel, E.; Bearden, D.; Rohr, J.;
Bechthold, A. Chem. Biol. 2000, 7, 821-831.
(28) Szu, P.-h.; He, X.; Zhao, L.; Liu, H.-w. Angew. Chem., Int. Ed. 2005, 44,
6742-6746.
(29) Chen, H.; Yeung, S.-M.; Que, N. L. S.; Muller, T.; Schmidt, R. R.; Liu,
H.-w. J. Am. Chem. Soc. 1999, 121, 7166-7167.
(30) Zhao, Z.; Hong, L.; Liu, H.-w. J. Am. Chem. Soc. 2005, 127, 7692-7693.
(31) Hong, L.; Zhao, Z.; Liu, H.-w. J. Am. Chem. Soc. 2006, 128, 14262-
14263.
(35) Smith, P.; Lin, A.; Szu, P.-h.; Liu, H.-w.; Tsai, S.-C. Acta Crystallogr.,
Sect. F: Struct. Biol. Cryst. Commun. 2006, 62, 231-234.
(36) Kieser, T.; Bibb, M. J.; Buttner, M. J.; Chater, K. F.; Hopwood, D. A.
Practical Streptomyces Genetics; The John Innes Foundation: Norwich,
England, 2000.
(37) Bradford, M. M. Anal. Biochem. 1976, 72, 248-254.
(38) Laemmli, U. K. Nature 1970, 227, 680-685.
(32) Chang, C.; Zhao, L.; Yamase, H.; Liu, H-w. Angew. Chem. 2000, 39, 2160-
2163.
(33) Chen, H.; Guo, Z.; Liu, H.-w. J. Am. Chem. Soc. 1998, 120, 9951-9952.
(34) Chen, H.; Yamase, H.; Murakami, K.; Chang, C. W.; Zhao, L.; Zhao, Z.;
Liu, H.-w. Biochemistry 2002, 41, 9165-9183.
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4957

A R T I C L E S
Hong et al.
Gene Amplification and Cloning of spnO, N, Q, R, and S. Genes
spnO, spnN, spnQ, spnR, and spnS were each PCR-amplified from S.
spinosa genomic DNA using primers with engineered 5′ NdeI and 3′
XhoI (HindIII in the case of spnO) restriction sites. The primer used
were the following, spnO: O-1 (5′-GGGGCCTCCATATGAGCAGT-
TCTGTCGAAGC-3′) and O-2 (5′-GAGAAGCTTTTATCG-CCCC-
AACGCCCACAAG-3′), spnN: N-1 (5′-GCGGCATATGACCAGCTC-
GATGCGAA-3′) and N-2 (5′-GGAGCTCGAGTGTGGACCCGCA-
3′), spnQ: Q-1 (5′-GGCCATATGCAGAGCCGGAAAACCAGA-3′)
and Q-2 (5′-GCCCTCGAGCTAGGAACTCTTGGCCAC-3′), spnR:
R-1 (5′-GCCCATATGATCAACCTGCACCA-3′) and R-2 (5′-GC-
CCTCGAGCTTTCGGAGTG-3′), and spnS: S-1 (5′-GGG-AATTC-
CATATGTCGCGCGTGAGCGACAC-3′) and S-2 (5′-ATCCGCTC-
GAGGTTGCGGATTCCGACGAAC-3′). Engineered restriction sites
are shown in bold, start codons in bold underlined, and stop codons
underlined. The PCR-amplified genes were purified, digested with the
appropriate restriction enzymes, and ligated into the appropriate vector
(pET24b+ for spnN, spnQ, spnR, and spnS, and pET28b+ for spnO)
digested with the same enzymes. After transformation, positive clones
were confirmed by DNA sequencing. The resulting plasmids were used
to transform E. coli BL21(DE3) for protein overexpression. The
recombinant SpnO is N-terminal His6-tagged, while SpnN, SpnR, and
SpnS are C-terminal His6-tagged. The expressed SpnQ protein carries
no tag.
Lysate and beads were then loaded onto a column, which was allowed
to drain and then washed with 25 mL of each of four wash buffers (50
mM NaH2PO4, 300 mM NaCl, 15% glycerol, pH 8.0) containing
incrementally increasing concentrations of imidazole (40, 60, 80, and
100 mM). Bound protein was eluted using 10.5 mL elution buffer (50
mM NaH2PO4, 300 mM NaCl, 250 mM imidazole, 15% glycerol, pH
8.0) and collected in 7 × 1.5 mL portions, which were pooled and
dialyzed against 4 × 1 L of 20 mM Tris·HCl, pH 7.5, 15% glycerol.
SpnO, SpnR, and SpnS eluted with buffer containing 250 mM
imidazole, while the purest fractions of SpnN eluted with wash buffer
containing 100 mM imidazole. After the Ni-NTA affinity chromato-
graphic step, SpnR and SpnS were further purified by a Sephacryl S-200
HR column (2.5 × 100 cm) using 20 mM Tris·HCl, pH 7.5, with a
flow rate of 1 mL/min. Proteins were concentrated by ultrafiltration
using an Amicon concentrator with an YM-10 membrane (Amicon)
and stored at -80 °C.
Preparation of Reductase Pairs Ferredoxin-Ferredoxin Reduc-
tase and Flavodoxin (FLD)-Flavodoxin Reductase (FDR). Ferredoxin
and ferredoxin reductase used in these experiments were commercial
products. The flavodoxin and flavodoxin reductase genes were amplified
from E. coli genomic DNA by PCR and cloned into pET24b(+) to
give proteins with C-terminal His6 tags upon expression. The growth
of E. coli BL21(DE3)-fld/pET24b(+) and E. coli BL21(DE3)-fdr/
pET24b(+) and purification of the recombinant FLD and FDR followed
similar procedures to those described above for forosamine biosynthetic
enzymes with the following modifications. Growth media (2 × 1 L)
were inoculated with overnight cultures (4 mL/L culture). The cultures
were induced at OD600 of 0.3 with 0.1 mM IPTG and grown for 6 h at
37 °C before harvesting. All buffers for Ni-NTA chromatography
contained 5 mM 2-mercaptoethanol. Dialysis buffer consisted of 50
mM Tris·HCl, pH 7.5, 15% glycerol.
Codon Changes in the spnN Gene. After the plasmid spnN/
pET24b+ was obtained, the AGGAGG sequence in spnN encoding
Arg-Arg (nucleotide 352-357) was mutated to CGTCGC, using the
QuikChange Site-Directed Mutagenesis Kit according to the recom-
mended procedures. The two complementary primers used for the
mutagenesis were 5′-GGGCTGGCCCGTCGCAAGAACCTG-3′ and
5′-CAGGTTCTTGCGACGGGCCAGC-3′. The resulting plasmid was
referred to as spnN/pET24b+-mut.
Iron Quantitation of SpnQ Protein. Iron quantitation was per-
formed according to the method of Fish.40 Briefly, a 1 mL solution
containing as-isolated SpnQ (1 mg) was treated with 0.5 mL of freshly
prepared reagent A (2.25% w/v KMnO4 in 6 N HCl) at 60 °C for 2 h.
After the treatment was complete, 100 µL of freshly prepared reagent
B (6.5 mM ferrozine, 13.1 mM neocuprine, 2 M ascorbic acid, and 5
M ammonium acetate in 25 mL ddH2O) was added. The mixture was
vigorously mixed and then incubated for 30 min at room temperature.
Standards made using different concentrations of Fe(NH4)2(SO4)2
solution in 0.01 M HCl were treated in parallel with the protein sample,
and the iron content of SpnQ was determined by comparing the
absorbance of the SpnQ sample at A562 to the standard curve.
Estimation of the PLP Content of SpnR and the Reconstitution
of SpnR with PLP. The PLP content of SpnR was estimated according
to literature procedures.41 Reconstitution of SpnR was carried out as
follows. A 200 µL solution containing SpnR (120 µM), PLP (2.0 mM),
DTT (1 mM), and 50 mM KH2PO4 buffer (pH 7.5) was incubated at
room temperature for 6 h. The entire solution was then loaded onto an
Econo-Pac 10 DG column. The reconstituted protein was eluted with
50 mM KH2PO4 buffer (pH 7.5). Fractions containing the SpnR protein
were pooled and concentrated. Any residual unbound PLP was removed
by repeated dilution and concentration using Centricon YM-10 micro-
concentrator.
Growth of E. coli SpnO, SpnN, SpnQ, SpnR, and SpnS
Recombinant Strains. An overnight culture of E. coli BL21(DE3)-
spnO/pET28b(+) strain grown in Luria-Bertani (LB) medium contain-
ing 50 µg/mL kanamycin at 37 °C was used (1 mL each) to inoculate
6 × 1 L of the same growth medium. These cultures were incubated
at 37 °C with vigorous shaking until the OD600 reached 0.4. Cultures
were precooled to 24 °C prior to induction. Protein expression was
then induced by the addition of isopropyl â-D-thiogalactopyranoside
(IPTG) to a final concentration of 0.2 mM, and the cells were allowed
to grow at 24 °C for an additional 15 h. The cells were harvested by
centrifugation at 4500g for 15 min and stored at -80 °C until lysis.
The growth of E. coli BL21(DE3)-spnQ/pET24b(+) and BL21(DE3)-
spnR/pET24b(+) strains followed an identical procedure. The growth
of E. coli BL21(DE3)-spnS/pET24b(+) followed a similar procedure,
except IPTG (0.1 mM) induction was carried out on cultures precooled
to 30 °C after OD600 reaching 0.4, and cultures were grown at 30 °C
for 7 h after induction. The growth of E. coli BL21(DE3)-spnN/pET24b-
(+)-mut also followed a similar procedure, except IPTG (0.25 mM)
induction was carried out on cultures precooled to 14 °C after OD600
reaching 0.6, and cultures were grown at 14 °C for 48 h after induction.
Purification of N-His6-Tagged SpnO, SpnN, SpnR, SpnS, and
Native SpnQ. The protocol for the purification of SpnQ was previously
reported.31 All purification steps of the His6-tagged proteins were carried
out at 4 °C. Thawed cells were first resuspended in lysis buffer (50
mM NaH2PO4, 300 mM NaCl, 5 mM imidazole, 15% glycerol, pH
8.0; 2 mL/g cells). Cells were then disrupted by sonication using 5 ×
1 min pulses with 1 min pauses between each pulse. The lysate was
centrifuged at 30 000g for 25 min, and the supernatant was subjected
to Ni-NTA chromatography according to the manufacturer’s protocol
with minor modifications. Briefly, the soluble protein fraction was
incubated with washed Ni-NTA beads on a rotator at 4 °C for 1 h.
Molecular Mass Determination. The native molecular masses of
proteins were determined by gel filtration chromatography using an
FPLC equipped with a Superdex 200 HR 10/30 column. The proteins
were eluted using 50 mM NaH2PO4, pH 7.0, 150 mM NaCl at a flow
rate of 0.8 mL/min. Calibration of the column (Vi) was achieved using
protein standards (Sigma). The void volume (Vo) of the column was
measured using blue dextran. The data were analyzed by the method
of Andrews.42
(40) Fish, W. W. Methods Enzymol. 1988, 158, 357-364.
(41) Roy, M.; Miles, E. W.; Phillips, R. S.; Dunn, M. F. Biochemistry 1988,
27, 8661-8669.
(39) Sambrook, J.; Russell, D. W. Molecular Cloning: A Laboratory Manual,
3rd ed.; Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY,
2001.
(42) Andrews, P. Biochem. J. 1964, 91, 222-233.
9
4958 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

TDP-D-Forosamine Biosynthesis in the Spinosyn Pathway
A R T I C L E S
Scheme 4
Synthesis of TDP-D-glucose (9). Compound 9 was prepared by
chemical coupling of R-D-glucose-1-phosphate (8) and TMP morpholi-
date. Thymidine monophosphate (TMP) and compound 8 were
purchased as sodium salts and were converted to their acidic forms
using Dowex cation exchange resin. The acidic form of compound 8
was then converted to a triethylamine salt and dried by lyophilization.
The acidic form of TMP was converted to TMP-morpholidate according
to the procedure of Moffatt and Khorana,43 affording the 4-morpholine
N,N′-dicyclohexylcarboxamidine salt of TMP-morpholidate in 90%
yield. 1H-tetrazole was purchased as a 3% (w/v) solution in acetonitrile.
Evaporation in Vacuo gave the solid white powder 1H-tetrazole.
Glucose-1-phosphate triethylamine salt (540 mg, 1.5 mmol) and TMP-
morpholidate (1.54 g, 2.25 mmol) were dissolved in 8 mL of anhydrous
pyridine, and the solution was evaporated to dryness in Vacuo. The
process of dissolving and evaporation was repeated twice with 1H-
tetrazole (190 mg, 2.7 mmol) added to the solution the second time.
After all water had been removed in this process, the mixture was
dissolved in 12 mL of anhydrous pyridine and stirred at room
temperature under nitrogen for 3 days. The solvent was then removed
by evaporation in Vacuo, and the residue was resuspended in 20 mL of
water and extracted with chloroform (3 × 15 mL). The aqueous portion
was dried by lyophilization, and the solid residue was redissolved in 2
mL of water and applied to a Bio-Rad P2 (extra fine) column (2.5 ×
120 cm). The column was run with 25 mM NH4HCO3 buffer at a flow
rate of 0.24 mL/min, and 6 mL fractions were collected. Fractions
exhibiting UV absorption at 267 nm were lyophilized, and the identity
1
and purity of the compounds in each fraction were determined by H
displaying a single absorption maximum at 267 nm were concentrated
and analyzed by NMR. Fractions containing 12 of high purity were
pooled and concentrated by partial lyophilization. The concentration
of 12 was determined spectrophotometrically based on a molar
extinction coefficient (ꢀ) of 9600 M-1 cm-1 at 267 nm for thymidine.45
and 31P NMR spectroscopy. Fractions containing significant quantities
of 9 were further purified by 1-2 passages through a Dowex cation
exchange column using water as the solvent. These fractions were
concentrated to 12 mg/mL. The fractions were neutralized with
ammonium bicarbonate and lyophilized individually, and their purity
was assessed by 1H and 31P NMR. The yield of 9 was 310 mg (>90%
pure, 37% yield). 1H NMR spectrum of 9 was identical to that
previously reported.44
Enzymatic Preparation of TDP-4-keto-6-deoxy-D-glucose (10).
Compound 10 was prepared from 9 by an enzymatic reaction using
RfbB. Preparation of RfbB enzyme followed a previously reported
procedure.44 The reaction mixture (1.5 mL) contained 9 (30 mg, 35
mM) and RfbB (24 µM) in 20 mM Tris·HCl buffer (pH 7.5). The
reaction was carried out at 37 °C for 2 h. The RfbB protein was removed
with Centricon YM-10 microconcentrator (Amicon), and the filtrate
containing 10 (25 mg, 78% yield, >90% pure) was used directly without
further purification. 1H NMR spectrum of 10 was identical to that
previously reported.46
Enzymatic Preparation of TDP-4-keto-2,6-dideoxy-D-glucose (12).
Compound 12 was prepared from 9 using RfbB, TDP-4-keto-6-deoxy-
D-glucose 2,3-dehydratase (TylX3) from the tylosin biosynthetic
pathway of S. fradiae19 and the expressed TDP-3,4-diketo-2,6-dideoxy-
D-glucose 3-ketoreductase, SpnN. The reaction mixture (1.5 mL)
contained 9 (25 mg, 30 mM), RfbB (53 µM), NADPH (34 mM), TylX3
(10 µM), and SpnN (17 µM) in 50 mM Tris·HCl buffer (pH 7.5)
containing 10% glycerol. The reaction was incubated at 24 °C, and its
progress was monitored spectrophotometrically by following the
consumption of NADPH at 340 nm. After the decrease of the
absorbance at 340 nm stopped (after 7 h), the enzymes were removed
with a Centricon YM-10 microconcentrator. The filtrate was applied
to a Bio-Rad P2 gel filtration column (extra fine, 2.5 × 120 cm) which
was then eluted with 25 mM NH4HCO3 at a flow rate of 0.24 mL/min.
Fractions (6 mL each) were analyzed spectrophotometrically, and those
1
The yield of 12 was 10 mg (>90% pure, 38% yield). H NMR of 12
(D2O) δ 1.07 (3H, d, J ) 6.3 Hz, 5-Me), 1.71 (1H, m, 2-Hax), 1.79
(3H, s, 5′′-Me), 2.01 (1H, m, 2-Heq), 2.22 (2H, m, 2′-H), 3.88 (1 H,
dd, J ) 11.7, 5.1 Hz, 3-H), 3.90 (1H, q, J ) 6.3 Hz, 5-H), 4.03 (3H,
m, 4′-H, 5′-Hs), 4.48 (1H, m, 3′-H), 5.48 (1H, dd, J ) 6.9, 3.3 Hz,
1-H), 6.21 (1H, t, J ) 7.2 Hz, 1′-H), 7.62 (1H, s, 6′′-H). 13C NMR
(D2O) δ 11.6, 11.9, 36.1 (d, J ) 6.3 Hz), 38.9, 68.3, 70.3, 71.2, 85.2,
85.6 (d, J ) 9.3 Hz), 93.3, 94.7 (d, J ) 5.5 Hz), 112.0, 137.6, 152.0,
166.9, 207.2. 31P NMR (D2O) δ -10.7 (d, J ) 21.2 Hz), -12.7 (d, J
) 21.2 Hz). HRMS (FAB) of 12, calcd for C16H23N2O14P2 [M - H]-
529.0625, found 529.0627.
Chemical Synthesis of TDP-C-4-Amino-2,3,4,6-tetradeoxy-r-D-
erythro-hexopyranose (24) (Scheme 4). (5R,6R)-6-Benzyloxy-5-
hydroxy-L-heptene (29). To a solution of aldehyde 28 (20.0 g, 121
mmol) in diethyl ether (100 mL) at 0 °C was added dropwise over 30
min a solution of 3-butenylmagnesium bromide in dry THF (130 mL).
The solution was stirred at room temperature for 3 h and then poured
into saturated ammonium chloride solution (300 mL). The resulting
two-phase mixture was separated. The aqueous layer was washed twice
with ether, and the combined organic layers were washed with brine,
dried over anhydrous Na2SO4, and concentrated in vacuo to give an
oily residue. Purification of the crude product by flash column
chromatography on silica gel using 4:1 hexanes-EtOAc afforded 15.2
g (57%) of 29, and 2.5 g (9%) of the minor diastereomer. Data for 29:
1H NMR (300 MHz, CDCl3) δ 7.39 (5H, m), 5.90 (1H, ddt, J ) 17.1,
10.2, 6.6 Hz), 5.10 (1H, d, J ) 17.1 Hz), 5.02 (1H, d, J ) 10.2 Hz),
4.71 (1H, d, J ) 11.4 Hz), 4.49 (1H, d, J ) 11.4 Hz), 3.51 (1H, m),
3.44 (1H, dq, J ) 6.3, 6.0 Hz), 2.73 (1H, d, J ) 3.6 Hz), 2.31 (1H, m),
2.21 (1H, m), 1.60 (2H, m), 1.24 (3H, d, J ) 6.0 Hz); 13C NMR (75
MHz, CDCl3) δ 138.8, 138.6, 128.7, 128.1, 128.0, 115.0, 78.6, 74.5,
71.2, 32.5, 30.2, 15.8; HRMS (CI) calcd for C14H21O2 [M + H]+
221.1541, found 221.1531.
(43) Moffatt, J. G.; Khorana, H. G. J. Am. Chem. Soc. 1961, 83, 649-658.
(44) Melanc¸on, C. E., III; Hong, L.; White, J. A.; Liu, Y. N.; Liu, H.-w.
Biochemistry 2007, 46, 577-590.
(45) Beaven, G. H.; Holiday, E. R.; Johnson, E. A. Nucleic Acids; Academic
Press: New York: 1955; Vol. 1.
(46) Cox, R. J.; Sherwin, W. A.; Lam, L. K. P.; Vederas, J. C. J. Am. Chem.
Soc. 1996, 118, 7449-7460.
(5S,6R)-5-Azido-6-benzyloxy-L-heptene (30). To a solution of
alcohol 29 (13.5 g, 58 mmol), triphenylphosphine (23.0 g, 87 mmol),
and diisopropyl azodicarboxylate (18.0 mL, 87 mmol) in dry THF (200
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4959

A R T I C L E S
Hong et al.
mL) was added dropwise diphenylphosphoryl azide (25.0 g, 87 mmol)
at 0 °C. The resulting mixture was concentrated after stirring at room
temperature for 12 h. The residue was quickly passed through a short
silica gel column and washed with 5:1 hexanes-EtOAc. The eluate
was concentrated under reduced pressure and purified by silica gel flash
column chromatography (10:1 hexanes-EtOAc) to give 8.3 g (55%)
of azide 30 as a colorless oil. 1H NMR (300 MHz, CDCl3) δ 7.39 (5H,
m), 5.84 (1H, m), 5.11 (1H, m), 5.07 (1H, m), 4.66 (1H, d, J ) 11.7
Hz), 4.58 (1H, d, J ) 11.7 Hz), 3.64 (1H, m), 3.51 (1H, m), 2.29 (1H,
m), 2.18 (1H, m), 1.62 (2H, m), 1.28 (3H, d, J ) 6.3 Hz); 13C NMR
(75 MHz, CDCl3) δ 138.4, 137.6, 128.7, 128.0, 127.8, 115.9, 77.4,
71.1, 65.5, 30.9, 29.7, 15.4; HRMS (CI) calcd for C14H20N3O [M +
H]+ 246.1606, found 246.1615.
sodium bicarbonate, dried over anhydrous Na2SO4, and evaporated in
vacuo. The residue was flash chromatographed on silica gel (hexanes-
EtOAc: 10:1 to 3:1) to give 8.5 g of the C-6 deprotected alcohol.
To a solution of the C-6 deprotected alcohol (8.4 g, 14.4 mmol) in
dry benzene (300 mL) were added 95% NaH (800 mg) and dimethyl
sulfoxide (2 mL). After the mixture was stirred at 80 °C for 1 h, it was
cooled to room temperature and filtered through a short silica gel pad.
The filtrate was concentrated in vacuo, and the residue was purified
by silica gel flash column chromatography (hexanes-EtOAc: 10:1 to
4:1) to give 4.4 g (66% yield, two steps) of the C-glycoside 33 as a
1
colorless oil. H NMR (300 MHz, CDCl3) δ 7.70 (4H, m), 7.44 (6H,
m), 3.90 (1H, m), 3.81 (1H, dd, J ) 11.2, 5.7 Hz), 3.72 (1H, dd, J )
11.2, 6.9 Hz), 3.63 (1H, dq, J ) 6.6, 6.3 Hz), 3.15 (1H, m), 1.96 (1H,
m), 1.79 (2H, m), 1.68 (1H, m), 1.24 (3H, d, J ) 6.3 Hz), 1.10 (9H,
s); 13C NMR (75 MHz, CDCl3) δ 135.9, 133.7, 130.0, 128.0, 71.5,
70.6, 64.4, 61.8, 27.1, 24.1, 24.0, 19.5, 18.4; HRMS (CI) calcd for
C23H30N3O2Si [M + H]+ 408.2107, found 408.2098.
(2R,5S,6R)-5-Azido-6-benzyloxyheptane-1,2-diol (31). A suspen-
sion of compound 30 (8.9 g, 87 mmol) and AD-mix-â (45 g) in tert-
butanol (150 mL) and water (150 mL) was vigorously stirred for 17 h
at 0-4 °C. The mixture was diluted with EtOAc (200 mL) and
quenched with saturated Na2S2O3 to give a purple solution. The organic
layer was separated, washed with brine, dried over anhydrous Na2-
SO4, and evaporated in vacuo. The residue was purified by silica gel
flash column chromatography (hexanes-EtOAc: 2:1 to 1:1) to afford
1-tert-Butyldiphenylsilyloxymethyl-4-N-carbo-tert-butoxyamino-
1,2,3,4,6-pentadeoxy-r-D-erythro-hexopyranose (34). To a solution
of the C-glycoside 33 (1.7 g, 4.0 mmol) in EtOAc (40 mL) were added
Boc2O (1.6 g, 7.2 mmol) and 10% Pd-C (350 mg) at room temperature.
The suspension was flashed with H2, and stirred under H2 atmosphere
for 5 h. The reaction mixture was then filtered through a short silica
gel pad. The filtrate was concentrated in vacuo, and the residue was
purified by silica gel flash column chromatography (hexanes-EtOAc:
10:1 to 5:1) to give 1.75 g (87%) of the product 34 as a viscous oil. 1H
NMR (400 MHz, CDCl3) δ 7.66 (4H, m), 7.40 (6H, m), 4.97 (1H, d,
J ) 8.8 Hz), 3.77 (2H, m), 3.68 (1H, dd, J ) 10.4, 5.2 Hz), 3.59 (1H,
dd, J ) 10.4, 6.0 Hz), 3.43 (1H, m), 1.84 (1H, m), 1.58 (3H, m), 1.44
(9H, s), 1.24 (3H, d, J ) 7.2 Hz), 1.06 (9H, s); 13C NMR (100 MHz,
CDCl3) δ 155.5, 135.9, 133.8, 129.9, 127.9, 79.4, 73.1, 69.9, 66.4, 49.2,
28.6, 27.1, 23.5, 19.5, 17.0; HRMS (CI) calcd for C28H42NO4Si [M +
H]+ 484.2883, found 484.2867.
1
5.5 g (63%) of the diol (31) as a colorless oil. H NMR (300 MHz,
CDCl3) δ 7.36 (5H, m), 4.63 (1H, d, J ) 11.7 Hz), 4.53 (1H, d, J )
11.7 Hz), 3.63 (3H, m), 3.45 (2H, m), 3.13 (1H, m), 3.00 (1H, s), 1.71
(1H, m), 1.63 (1H, m), 1.45 (2H, m), 1.24 (3H, d, J ) 6.3 Hz); 13C
NMR (75 MHz, CDCl3) δ 138.4, 128.7, 128.0, 127.9, 77.5, 72.4, 71.1,
66.9, 66.5, 30.4, 26.8, 15.3; HRMS (CI) calcd for C14H22N3O3 [M +
H]+ 280.1661, found 280.1652.
(2R,5S,6R)-5-Azido-6-benzyloxy-L-(tert-butyldiphenylsilyloxy)-2-
heptyl Tosylate (32). To a solution of diol 31 (5.5 g, 19.7 mmol) and
tert-butyldimethylsilyl chloride (5.8 g, 21 mmol) in dichloromethane
(150 mL) was added imidazole (1.63 g, 24 mmol) at 0 °C. After being
stirred for 2 h at room temperature, the mixture was diluted with
dichloromethane, washed with brine, dried over anhydrous Na2SO4,
and evaporated in vacuo. The residue was separated by silica gel flash
chromatography to give 9.7 g of the silyl ether intermediate.
Diethyl (2S,5S,6R)-(5-N-Carbo-tert-butoxyamino-6-methyltet-
rahydropyranyl)-2-methanephosphonate (35). A solution of 34 (1.75
g, 3.6 mmol) and tetrabutylammonium fluoride (5 mmol) in dry THF
(30 mL) was stirred at room temperature for 7 h and then concentrated
in vacuo. The residue was diluted with diethyl ether, washed with water,
dried over anhydrous Na2SO4, evaporated in vacuo, and purified by
silica gel flash column chromatography (hexanes-EtOAc: 2:1 to 1:2)
to give 0.82 g (92%) of the deprotected alcohol as a colorless oil.
To a solution of the deprotected alcohol (0.81 g, 3.3 mmol) and
triphenylphosphine (1.73 g, 6.6 mmol) was added portionwise carbon
tetrabromide (2.2 g, 6.6 mmol) at room temperature. The mixture was
stirred for 11 h and then concentrated. The residue was separated by
silica gel flash chromatography (hexanes-EtOAc: 10:1 to 3:1) to give
0.58 g (56%) of the corresponding bromide as a colorless oil.
A mixture of the above bromide (0.58 g, 1.8 mmol), triethyl
phosphite (6.0 mL), and tetrabutylammonium iodide (50 mg) was stirred
at 120 °C for 42 h under N2, and then evaporated over an oil pump.
The residue was purified by silica gel flash column chromatography
(hexanes-EtOAc: 3:1 to 1:2 plus 5% ethanol) to give 0.59 g (85%)
Triethylamine (3.9 mL, 28 mmol) was added dropwise to a solution
of the intermediate (9.7 g, 18.7 mmol) in dichloromethane (140 mL)
containing tosyl chloride (4.5 g, 23 mmol) and 4-dimethylaminopyridine
(100 mg). The solution was stirred at room temperature for 14 h, at
which time additional tosyl chloride (2.0 g, 10 mmol) and triethylamine
(3.0 mL, 21 mmol) were introduced, and the reaction was continued
for another 12 h. The mixture was diluted with diethyl ether, washed
sequentially with saturated sodium bicarbonate, 1 N HCl, and brine,
dried over anhydrous Na2SO4, and evaporated in vacuo. The residue
was purified by silica gel flash column chromatography (hexanes-
EtOAc: 10:1 to 3:1) to give 9.8 g (74% yield, two steps) of the
1
protected diol 32 as a yellowish oil. H NMR (300 MHz, CDCl3) δ
7.73 (2H, d, J ) 8.4 Hz), 7.62 (4H, m), 7.39 (11H, m), 7.24 (2H, d, J
) 8.4 Hz), 4.62 (1H, d, J ) 11.7 Hz), 4.55 (1H, m), 4.52 (1H, d, J )
11.7 Hz), 3.67 (2H, m), 3.53 (1H, m), 3.36 (1H, dt, J ) 10.2, 3.6 Hz),
2.40 (3H, s), 1.97 (1H, m), 1.72 (1H, m), 1.45 (2H, m), 1.16 (3H, d,
J ) 6.0 Hz), 1.06 (9H, s); 13C NMR (75 MHz, CDCl3) δ 144.9, 138.4,
135.9, 135.8, 134.4, 133.2, 133.0, 130.1, 130.0, 128.7, 128.0, 127.9,
127.8, 127.7, 82.9, 77.4, 71.0, 66.0, 64.8, 28.7, 28.0, 27.0, 26.2, 21.9,
19.5, 15.1; HRMS (CI) calcd for C37H46N3O5SiS [M + H]+ 672.2927,
found 672.2931.
1
of the desired phosphonate 35 as a colorless oil. H NMR (300 MHz,
CDCl3) δ 5.04 (1H, d, J ) 9.0 Hz), 4.09 (5H, m), 3.78 (1H, m), 3.42
(1H, m), 1.85-2.13 (4H, m), 1.57-1.74 (2H, m), 1.43 (9H, s), 1.29
(9H, m); 13C NMR (75 MHz, CDCl3) δ 155.5, 79.5, 73.0, 65.3, 61.9
(d, J ) 10.0 Hz), 61.8 (d, J ) 10.0 Hz), 49.1, 28.6, 27.9 (d, J ) 15.6
Hz), 23.9, 17.1, 16.7, 16.6; 31P NMR (121 MHz, CDCl3) δ 29.7; HRMS
(CI) calcd for C16H33NO6P [M + H]+ 366.2046, found 366.2052.
C-(4-Amino-1,2,3,4,6-pentadeoxy-r-D-erythro-hexopyranosyl)-
methanephosphonic Acid (36). To a solution of the phosphonate 35
(0.58 g, 1.5 mmol) and pyridine (0.8 mL) in dry acetonitrile (23 mL)
was added slowly trimethylsilyl bromide (2.2 mL). The mixture was
stirred for 35 h at room temperature and then concentrated in vacuo.
The oily residue was treated with water (20 mL) in the presence of
excess ammonium bicarbonate. The aqueous solution was extracted
4-Azido-1-tert-butyldiphenylsilyloxymethyl-1,2,3,4,6-pentadeoxy-
r-D-erythro-hexopyranose (33). To a solution of the protected diol
32 (10.8 g, 16.0 mmol) in anhydrous dichloromethane (180 mL) was
added dropwise 1.0 M BCl3 (50 mL) in dichloromethane at -78 °C
under N2. The solution was stirred for 4 h at the same temperature, at
which time TLC indicated complete consumption of the starting
material. The reaction was quenched by dropwise addition of methanol
(8.0 mL) and triethylamine (11.0 mL). The solution was warmed to
room temperature, diluted with diethyl ether, washed with saturated
9
4960 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

TDP-D-Forosamine Biosynthesis in the Spinosyn Pathway
A R T I C L E S
with chloroform (10 mL) and then lyophilized to give the crude product
as a white powder. The crude product was dissolved in water, loaded
on a DOWEX 50W × 8 (H+ form) column (1.5 × 10 cm), washed
with water, and then with 1 M aqueous ammonia. The desired fractions
were combined and lyophilized to give 0.32 g (89%) of the phosphonic
acid product 36 as a white powder. 1H NMR (300 MHz, D2O, pD 7.5)
δ 4.02 (1H, m), 3.85 (1H, dq, J ) 6.3, 4.5 Hz), 3.06 (1H, m), 2.00
(1H, m), 1.78 (1H, m), 1.70 (3H, m), 1.53 (1H, m), 1.20 (3H, d, J )
6.9 Hz); 13C NMR (75 MHz, D2O, pD 7.5) δ 69.5, 67.8, 50.4, 33.8 (d,
J ) 128.6 Hz), 26.1 (d, J ) 4.4 Hz), 21.4, 15.9; 31P NMR (121 MHz,
D2O, pD 7.5) δ 19.7.
TylX3 (1.0 µM), SpnN (2.0 µM), and 50 mM KH2PO4 buffer (pH 7.5).
The reaction, carried out at 24 °C, was initiated by the addition of
TylX3, and was monitored by HPLC. The retention times were 34.5
min for TDP-D-glucose (9) and 33.4 min for TDP-4-keto-2,6-dideoxy-
D-glucose (12).
Activity Assays for SpnQ. A typical SpnQ assay mixture (100 µL)
under the “ColD-like” conditions contained 12 (1 mM), PLP (250 µM),
SpnQ (10 µM), and L-glutamate (3 mM) in 50 mM KH2PO4 buffer
(pH 7.5), and was incubated at 37 °C. The HPLC retention time was
15.8 min for TDP-4-amino-2,4,6-trideoxy-D-glucose (23). The efficiency
of L-aspartate as the amino donor was examined by replacing L-
glutamate with L-aspartate of the same concentration in the assay. A
typical SpnQ assay mixture (100 µL) under the “E1-like” conditions
contained 12 (0.7 mM), PMP (250 µM), SpnQ (30 µM), and the
reducing agent/system being tested (such as 0.6 mM sodium dithionite
or 30 µM each of flavodoxin/flavodoxin reductase or ferredoxin/
ferredoxin reductase in the presence of 0.7 mM NADPH) in 50 mM
KH2PO4 buffer (pH 7.5) and was incubated at 24 °C. The HPLC
retention time was 36.2 min for TDP-4-keto-2,3,6-trideoxy-D-glucose
(13). SpnQ assays were conducted under aerobic conditions using as-
purified enzyme except when dithionite was used as the reductant, in
which case anaerobically reconstituted SpnQ was assayed under
anaerobic conditions.
Thymidine Diphosphate C-4-Amino-2,3,4,6-tetradeoxy-r-D-erythro-
hexopyranose (24). A mixture of the phosphonic acid 36 (0.32 g, 1.4
mmol) and TMP morpholidate (1.5 g, 2.1 mmol) in anhydrous pyridine
(6 mL) was evaporated three times using an oil pump. To the resulting
solid were added anhydrous pyridine (8 mL) and 1H-tetrazole (160
mg, 2.2 mmol). The mixture was stirred vigorously at room temperature
for 115 h. The solvents were then removed using an oil pump, and the
remaining solid was dissolved in water (∼3 mL). This aqueous solution
was loaded onto a Bio-Gel P2 (extra fine) column (2.0 × 120 cm) and
eluted with 25 mM ammonium bicarbonate. The desired fractions, as
indicated by their UV absorption at 267 nm, were pooled and
lyophilized to afford 0.38 g (52%) of the isostere product 24 as a white
powder. 1H NMR (300 MHz, 2H2O, pD 7.5) δ 7.61 (1H, s, 6′′-H), 6.21
(1H, dd, J ) 7.2, 6.6 Hz, 1′-H), 4.48 (1H, m, 3′-H), 4.06 (4H, m, 1-H,
4′-H, 5′-H), 3.85 (1H, dq, J ) 6.9, 4.5 Hz, 5-H), 3.04 (1H, m, 4-H),
2.23 (2H, m, 2′-H), 1.97 (2H, dd, J ) 18.9, 6.9 Hz, CH2-P), 1.79
(3H, s, 5′′-Me), 2.02-1.48 (4H, m, 2-H, 3-H), 1.20 (3H, d, J ) 6.9
Isolation and Characterization of SpnQ Products 13 and 23. The
reaction conditions, isolation, and characterization of the products 13
and 23 obtained from incubation of SpnQ with 12 under the “ColD-
like” and the “E1-like” conditions, respectively, were previously
reported.31
2
Hz, 5-Me); 13C NMR (75 MHz, H2O, pD 7.5) δ 166.9, 152.0, 137.6,
Determination of the Kinetic Parameters for SpnQ Using 12.
The steady-state kinetic parameters of the SpnQ-catalyzed reaction were
determined using a continuous assay in which NADPH consumption
was monitored. A series of reactions (100 µL) containing SpnQ (1.0
µM), PMP (40 µM), NADPH (150 µM), ferredoxin (10 µM), ferredoxin
reductase (10 µM), and various concentrations of 12 (10-200 µM) in
a total volume of 50 mM KH2PO4 buffer (pH 7.5) were prepared and
monitored spectrophotometrically. The reactions were conducted at
24 °C, and each initial reaction rate was determined as the initial slope
of the decrease in absorbance at 340 nm. The resulting data were fit to
the Michaelis-Menten equation by nonlinear regression using Grafit
to determine the kcat and Km values.
111.9, 85.5 (d, J ) 8.7 Hz), 85.1, 71.1, 69.7, 66.9, 65.4 (d, J ) 6.0
Hz), 50.4, 38.7, 33.1 (d, J ) 136.4 Hz), 26.2 (d, J ) 6.4 Hz), 21.5,
2
15.9, 11.8; 31P NMR (121 MHz, H2O, pD 7.5) δ 14.4 (d, J ) 26.0
Hz), -10.6 (d, J ) 27.4 Hz); HRMS (FAB) calcd for C17H28N3O11P2
[M - H]- 512.1199, found 512.1189.
HPLC Assay Methods. Unless otherwise specified, all HPLC assays
were conducted using a Dionex CarboPac PA1 analytical column (4
× 250 mm) with a CarboPac PA1 guard column (4 × 50 mm). A
suitable amount of reaction mixture was withdrawn at appropriate time
intervals, and the reaction was quenched by 10-fold dilution with water
and flash frozen in liquid nitrogen. Prior to HPLC analysis, the samples
were thawed and filtered through a Microcon YM-10 membrane to
remove enzyme. The filtrate was then injected into the HPLC column.
The sample was eluted with a gradient of water as solvent A and 500
mM NH4OAc (adjusted to pH 7.0 with aqueous NH3) as solvent B.
The gradient was ran from 5 to 20% B over 15 min, from 20 to 60%
B over 20 min, from 60 to 100% B over 2 min, with a 3 min wash at
100% B, and from 100 to 5% B over 5 min, followed by reequilibration
at 5% B for 15 min. The flow rate was 1 mL/min, and the detector
was set at 267 nm.
Activity Assays for SpnR Using 12, 13, 24, and 27. The reaction
mixture (50 µL) for the SpnR assay using 4-keto substrates 12 or 27
contained substrate (1 mM), L-glutamate (10 mM), PLP (40 µM), and
as-purified SpnR (10 µM) in 50 mM KH2PO4 buffer (pH 7.5). The
HPLC retention times were 33.4 min for 12, 15.8 min for 23, 32.6 min
for 27, and 6.8 min for 24. The contents of a typical reaction mixture
for assaying SpnR using 4-amino substrate mimic 24 were identical to
those used above, except that R-ketoglutarate (30 mM) was used in
place of L-glutamate. The competence of pyruvate as the cosubstrate
in the SpnR reaction using 24 as substrate was examined by replacing
R-ketoglutarate with pyruvate of the same concentration in the assay.
The effect of increasing PLP concentrations on the SpnR reaction using
24 as substrate was examined by maintaining the above assay conditions
but varying the concentration of PLP (0, 10, 20, 40, or 80 µM). Each
reaction was stopped at 3 min and analyzed by HPLC. Because of its
instability, the physiological SpnR substrate 13 was generated in situ
prior to the assay by the incubation of 12 (0.7 mM) with SpnQ (30
µM), ferredoxin/ferredoxin reductase (30 µM each), and NADPH (2
mM) in 50 mM KH2PO4, pH 7.5 at 24 °C for 3 h. Enzymes were
removed by filtration through a Microcon YM-10 membrane. To the
filtrate were added SpnR, PLP, L-glutamate, and MgCl2 to give a final
concentrations of 37 µM, 305 µM, 12.2 mM, and 1.22 mM, respectively.
The reaction was initiated by the addition of SpnR and incubated at
24 °C for an additional 2 h. The progress of the reaction was monitored
by HPLC. The HPLC retention time was 9.2 min for the SpnR product
14.
Activity Assays for SpnO and SpnN and Coupled SpnO/SpnN
Assay. A typical SpnO assay mixture (50 µL) contained 10 (30 µM),
SpnO (4.2 µM), and 50 mM KH2PO4 buffer (pH 7.5). The reaction,
carried out at 24 °C, was initiated by the addition of SpnO, and its
progress was monitored by following changes in peak integrations of
substrate and product by HPLC as described above. The retention times
were 35.3 min for TDP-4-keto-6-deoxy-D-glucose (10) and 2.4 min
for maltol (22). The activities of SpnO and SpnN were examined
spectrophotometrically together at 24 °C by following the change in
absorbance at 340 nm, which corresponds to NADPH consumption
(ꢀ340 ) 6220 M-1 cm-1) by SpnN. A typical assay mixture to determine
SpnO and SpnN activity (100 µL) contained 9 (17 µM), NADPH (18
µM), RfbB (40 µM), SpnO (1 µM), and SpnN (2 µM) in 50 mM KH2-
PO4 buffer, pH 7.5. The reaction was initiated by the addition of SpnO.
The activity of SpnN was also examined by HPLC using TylX3 as the
dehydratase. A typical coupled assay mixture to determine SpnN activity
(50 µL) contained 9 (0.42 mM), RfbB (40 µM), NADPH (150 µM),
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4961

A R T I C L E S
Hong et al.
Isolation and Characterization of the SpnR Reaction Products
23 and 27. The reaction conditions, isolation, and NMR characterization
of 27 obtained from the incubation of SpnR with 24 in the presence of
excess R-ketoglutarate were previously reported.30 The isolation and
mass spectrometric characterization of 23 obtained from incubation of
SpnR with 12 and L-glutamate had also been reported.30
Determination of the Keq Value for the SpnR-Catalyzed Tran-
samination of 24. A reaction (50 µL) containing aminosugar 24 (0.6
mM), R-ketoglutarate (0.6 mM), PLP (50 µM), and as-purified SpnR
(10 µM), in 50 mM KH2PO4, pH 7.5, was prepared and incubated at
24 °C for 12 h. Ratios of 24 and ketosugar 27 were determined based
on HPLC peak integration. Values were inserted into eq 1 to calculate
the equilibrium constant (Keq) for the reaction.
Figure 2. SDS-PAGE gels (12%) of purified SpnO (lane 1), SpnN (lane
2), SpnQ (lane 3), SpnR (lane 4), and SpnS (lane 5).
was monitored by HPLC. After 17 h, about 70% of the substrate had
been converted to products as judged by substrate peak integration (30%
of the substrate remained unreacted). The enzyme was removed with
a Centricon YM-10 microconcentrator, and the filtrate was applied to
a Bio-Rad P2 gel filtration column (extra fine, 2.5 × 120 cm) which
was eluted with 25 mM NH4HCO3 at a flow rate of 0.24 mL/min.
Fractions (6 mL each) were analyzed spectrophotometrically, and those
displaying a single absorption maximum at 267 nm were concentrated
and analyzed by NMR. Fractions containing 25 of high purity were
pooled and concentrated by lyophilization. The concentration of 25
was determined spectrophotometrically based on a molar extinction
coefficient (ꢀ) of 9600 M-1 cm-1 at 267 nm for thymidine. The yield
of 25 was 3 mg (>95% pure, 33% yield). 1H NMR of 25 (D2O) δ 1.32
(3H, d, J ) 4.2 Hz, 5-Me), 1.85 (3H, s, 5′′-Me), 1.44-2.16 (6H, m,
2-H, 3-H, CH2-P), 2.24-2.34 (2H, m, 2′-H), 2.66 (3H, s, NH-Me),
3.02 (1H, m, 4-H), 4.05-4.16 (5H, m, 4′-H, 5′-H, 1-H, 5-H), 4.54 (1H,
ddd, 3′-H), 6.27 (1H, t, J ) 3.9 Hz, 1′-H), 7.67 (1H, s, 6′′-H). 13C
NMR (D2O) δ 11.9, 15.4, 19.2, 26.0, 30.9, 34.2 (d, J ) 138.2), 38.8,
[24] {[R-KG]0 - [27]}
[27]2
[24] [R-KG]
[27] [L-glu]
Keq
)
)
(1)
Determination of the Kinetic Parameters for SpnR Catalysis
Using 27 as Substrate. The steady-state kinetic parameters of the SpnR-
catalyzed transamination of 27 were determined using a continuous
L-glutamate dehydrogenase-coupled assay in which NADPH consump-
tion was monitored.46 Specifically, a series of reactions (100 µL)
containing as-purified SpnR (1.0 µM), PLP (250 µM), L-glutamate (50
mM), NADPH (150 µM), NH4Cl (80 mM), 0.6 U of L-glutamate
dehydrogenase, and various concentrations of 27 (5-1400 µM) in 50
mM KH2PO4 buffer (pH 7.5) were prepared and monitored spectro-
photometrically. The reactions were conducted at 24 °C, and each initial
reaction rate was determined by following the initial slope of the
decrease in absorbance at 340 nm. The resulting data were fit to the
Michaelis-Menten equation by nonlinear regression using Grafit to
app
determine the kcat and Km values.
57.7, 65.4, 66.1, 68.9, 71.1, 85.2, 85.6, 112.2, 138.0, 152.0, 167.2. 31
P
Activity Assays for SpnS Using 24. A typical assay mixture (60
µL) contained substrate 24 (1 mM), S-adenosyl-L-methionine (SAM,
2.0 mM), MgCl2 (2.0 mM), DTT (2.0 mM), SpnS (10 µM), and 50
mM KH2PO4 (pH 7.5). The reaction was initiated by the addition of
SpnS and was carried out at 37 °C. The progress of the reaction was
monitored by HPLC (Dionex CarboPac PA1 column). The HPLC
retention time was 5.1 min for the SpnS monomethylated product 25,
which overlaps with an authentic sample of S-adenosyl-L-homocysteine
(SAH). The peak of the dimethylated product 26 overlaps with those
associated with SAM, SAH, and its degradation products (retention
times 2.6-5.5 min) and thus could not be directly observed. The SpnS
reaction was also monitored using an Adsorbosphere SAX column (5
µm, 4.6 × 250 mm) at 267 nm. A linear gradient from 0 to 20% Buffer
D (500 mM KH2PO4 buffer, pH 3.5) in Buffer C (50 mM KH2PO4
buffer, pH 3.5) over 20 min gave baseline separation between substrate
24 (having a retention time of 11.0 min) and monomethylated product
25 (having a retention time of 20.2 min). Unfortunately, the peak of
the dimethylated product 26 using the SAX column also overlaps with
those associated with SAM, SAH, and its degradation products
(retention times 2.3-4.2 min) and therefore could not be directly
detected. Hence, for SpnS methylation time course experiments, the
amount of 26 formation at a given time point was calculated by
subtracting the integrations of the remaining substrate and the monom-
ethylated product peaks from that of the substrate peak recorded at
time zero. The effect of increasing Mg2+ concentrations on SpnS
catalytic efficiency using 24 as substrate was examined under the above
assay conditions but varying the concentration of MgCl2 (100, 200,
400, and 800 µM, and 2.0 and 10 mM). Each reaction was stopped at
20 min and analyzed by HPLC (Dionex CarboPac PA1 column) to
determine the extent of substrate consumption.
NMR (D2O) δ 14.1 (d, J ) 27 Hz), -10.7 (d, J ) 27 Hz). HRMS
(FAB) calcd for C18H32N3O11P2 [M + H]+ 528.1512, found 528.1502.
The monomethylated product 25 of the SpnS reaction isolated from
the preparative scale reaction described above was used to examine
the formation of the dimethylated product 26. Assay conditions were
identical to those used for the SpnS activity assays described above,
except that 20 µM SpnS was used in the reaction. The reaction was
carried out at 37 °C for 12 h. The enzyme was removed using a
Centricon YM-10 microconcentrator, and the filtrate was subjected to
Dionex-HPLC analysis. As has been mentioned before, neither 25 nor
26 could be resolved from the SAM degradation products under the
HPLC conditions used. Thus, all peaks with retention times of 2.3-
4.2 min were collected. The compounds were lyophilized to dryness
and redissolved in water, and the resulting solution was submitted to
MS analysis. ESI-MS calculated for 26, C19H32N3O11P2 [M - H]-, was
540, and a mass of 540 was found.
Results
Purification and Characterization of SpnO and SpnN. In
order to assess the catalytic activities of the proposed TDP-4-
keto-6-deoxy-D-glucose 2,3-dehydratase, SpnO, and TDP-3,4-
diketo-2,6-dideoxy-D-glucose 3-ketoreductase, SpnN, several
overexpression constructs of spnO and spnN were made. After
testing the overexpression level and solubility of SpnO protein
derived from these constructs, spnO/pET28b(+), which produces
N-terminal His6-tagged SpnO, was identified as the one that
gives the best yield of soluble protein. However, SpnO-N-His6
was prone to precipitation, and inclusion of 20% glycerol in all
buffers during purification was important for minimizing
precipitation. The isolated yield of SpnO-N-His6 was 2 mg of
pure protein (Figure 2, lane 1) per 6 L of culture after a single
Ni-NTA chromatographic step. The purity was estimated to be
>90%.
Isolation and Characterization of the SpnS Products 25 and 26.
A preparative scale incubation was carried out in which 24 (8.5 mg,
8.25 mM), SpnS (80 µM), SAM (20 mM), MgCl2 (2 mM), and DTT
(2 mM) were incubated in 2 mL of 50 mM KH2PO4 buffer (pH 7.5).
The reaction was carried out at 37 °C, and the progress of the reaction
9
4962 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

TDP-D-Forosamine Biosynthesis in the Spinosyn Pathway
A R T I C L E S
Expression and purification of SpnN were more problematic.
After different vectors and hosts failed to yield any detectable
SpnN expression, the spnN gene sequence was reexamined,
revealing two potential problems: the start codon in the
published spnN sequence was likely misassigned, and spnN
contains tandem AGG (Arg) codons which are rarely found in
E. coli genome and are known to drastically reduce protein
expression level47 and promote frameshifting.48 Recloning of
spnN using the alternative start codon 12 nucleotides upstream
of the originally assigned start codon and site-directed mu-
tagenesis of the rare Arg codons resulted in a significant
improvement in the yield of C-terminal His6-tagged SpnN
protein, allowing the isolation of 4 mg of >90% pure SpnN-
C-His6 (Figure 2, lane 2) from a 6 L culture after a single Ni-
NTA chromatographic step. The monomeric molecular weights
of 55 kDa for SpnO and 38 kDa for SpnN estimated by SDS-
PAGE are in good agreement with the values of 56 551 Da and
37 732 Da predicted from their gene sequences. Gel filtration
chromatographic analysis revealed that the molecular masses
for the purified SpnO-N-His6 and SpnN-C-His6 are 96 kDa and
38 kDa, respectively, indicating that SpnO is a homodimeric
protein and SpnN is a monomeric protein. Neither protein shows
absorbance above 300 nm.
Catalytic Properties of SpnO and SpnN. SpnO shares
significant sequence identity with the characterized TDP-4-keto-
6-deoxy-D-glucose 2,3-dehydratases, TylX3, Gra27, and OleV,
which convert 10 to TDP-3,4-diketo-2,6-dideoxy-D-glucose (11),
and was therefore proposed to play a similar role in forosamine
biosynthesis. SpnN shares nearly 50% sequence identity with
Gra26 and OleW, the ketoreductases shown to convert 11 to
TDP-4-keto-2,6-dideoxy-D-glucose (12) in their respective
pathways but only shares 14% sequence identity with TylC1,
the reductase which converts 11 to TDP-4-keto-2,6-dideoxy-
D-allose (21), the C-3 epimer of 12.19,21 SpnN was thus proposed
to convert the SpnO product 11 to 12. To verify the function of
SpnO, compound 10 (Figure 3, trace b), the proposed substrate
of SpnO, was chemoenzymatically synthesized and was sub-
sequently incubated with SpnO. Previous assays using SpnO
homologues demonstrated that 11 is unstable, rapidly degrading
to TDP and maltol (22). As expected, HPLC analysis of the
SpnO reaction mixture showed the formation of a new peak
which has the same retention time (2.4 min) as an authenic
sample of 22.
Figure 3. Enzymatic reactions of forosamine biosynthetic enzymes. (a)
Purified chemically synthesized TDP-D-glucose (9); (b) incubation of 9 with
RfbB in 20 mM Tris·HCl buffer (pH 7.5) to form TDP-4-keto-6-deoxy-D-
glucose (10); (c) purified SpnN product TDP-4-keto-2,6-dideoxy-D-glucose
(12) obtained by incubation of 9 (30 mM) with RfbB (53 µM), NADPH
(34 mM), TylX3 (10 µM), and SpnN (17 µM) in 50 mM Tris·HCl buffer
(pH 7.5), 10% glycerol; (d) incubation of 12 with SpnQ (30 µM), PMP
(250 µM), ferredoxin (30 µM), ferredoxin reductase (30 µM), and NADPH
(0.7 mM) in 50 mM KH2PO4 (pH 7.5) to form TDP-4-keto-2,3,6-trideoxy-
D-glucose (13); (e) incubation of 12 with SpnQ (10 µM), PLP (250 µM),
and L-glutamate (3 mM) in the same buffer as in d to form TDP-4-amino-
2,4,6-trideoxy-D-glucose (23); (f) incubation of 12 under the same conditions
as in d followed by removal of enzyme and incubation with SpnR (37 µM),
PLP (305 µM), L-glutamate (12.2 mM), and MgCl2 (1.22 mM) to form
TDP-4-amino-2,3,4,6-tetradeoxy-D-glucose (14); (g) incubation of 12 (1
mM) with as-purified SpnR (10 µM), PLP (40 µM), and L-glutamate (10
mM) in the same buffer as in d to form 23.
which together convert 10 to 12. Since NADPH was consumed
2-fold faster than NADH under otherwise identical assay
conditions, NADPH is likely the physiological reductant for the
SpnN-catalyzed reaction. A similar preference was also found
for SpnN homologues Gra26 and OleW, which use NADPH
more effectively.21
It should be noted that SpnO is not as efficient a 2,3-
dehydratase as its homologue, TylX3.19 Therefore, in subsequent
reactions TylX3 was substituted for SpnO for the preparation
of 12. The comparative catalytic efficiencies of SpnO and TylX3
were not assessed quantitatively because of the instability of
the reaction product, 11. When a coupled TylX3/SpnN reaction
performed using 9, RfbB, and NADPH was monitored by
HPLC, a new peak (retention time of 33.4 min), which was
presumed to be the SpnN product 12, was detected. Figure 3,
trace c, shows the HPLC chromatogram of the purified SpnN
product 12 obtained from a preparative-scale incubation of 9
with RfbB, TylX3, SpnN, and NADPH. The 1H NMR spectrum
of the product was fully consistent with that previously reported
for TDP-4-keto-2,6-dideoxy-D-glucose (12).21 This data, com-
bined with high-resolution FAB-MS of 12, unambiguously
established that SpnN is the ketoreductase responsible for the
conversion of 11 to 12 in the forosamine biosynthetic pathway.
Substrate Specificity and Steady-State Kinetics of the
SpnQ-Catalyzed Reaction. SpnQ was overexpressed and
purified and its identity confirmed by N-terminal amino acid
sequencing. SpnQ, which has a predicted molecular weight of
50 361, appeared as a 51 kDa band on SDS-PAGE (Figure 2,
lane 3) and was found to be a dimer in solution (molecular mass
109.4 kDa). SpnQ was subsequently shown in Vitro to be the
3-dehydrase responsible for conversion of the SpnN product
TDP-4-keto-2,6-dideoxy-D-glucose (12) to TDP-4-keto-2,3,6-
trideoxy-D-glucose (13, retention time of 36.2 min in Figure 3,
Since reduction of 11 by SpnN will give a stable product,
12, a tandem reaction of SpnO and SpnN was carried out to
confirm the activities of SpnO as well as SpnN. Both enzymes
were incubated together with TDP-D-glucose (9), RfbB, and
either NADH or NADPH, and the reaction was monitored
spectrophotometrically by following the consumption of NAD-
(P)H at 340 nm. In this reaction, 9 was first converted to the
SpnO substrate 10 in situ using RfbB. A time-dependent
decrease in A340 was observed in reactions containing SpnO,
SpnN, and either NADH or NADPH, but no decrease in A340
was observed in control reactions lacking either SpnO or SpnN.
These results support the proposed roles of SpnO and SpnN as
the 2,3-dehydratase and the NAD(P)H-dependent ketoreductase
(47) Chumpolkulwong, N.; Sakamoto, K.; Hayashi, A.; Iraha, F.; Shinya, N.;
Matsuda, N.; Kiga, D.; Urushibata, A.; Shirouzu, M.; Oki, K.; Kigawa, T.;
Yokoyama, S. J. Struct. Funct. Genomics 2006, 7, 31-36.
(48) Gurvich, O. L.; Baranov, P. V.; Gesteland, R. F.; Atkins, J. F. J. Bacteriol.
2005, 187, 4023-4032.
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4963

A R T I C L E S
Hong et al.
Table 1. Summary of Kinetic Parameters of Reactions of
Characterized PMP-Dependent 3-Dehydrases SpnQ, E1, and ColD
1
1
enzyme
k
cat (min-
)
Km
(
µM)
k
cat/Km (min-1 µM-
)
SpnQ
E1
ColD
2.6
3.5
36
49
44
0.053
0.079
0.73a
49a
app
a The Km value for the ColD-catalyzed reaction is a Km
.
Figure 4. UV/visible spectrum of as-purified SpnQ (8.7 µM) showing
characteristic [2Fe-2S] features at 323, 458, and 560 nm.
trace d) in the forosamine biosynthetic pathway.31 SpnQ, like
its homologue E1 (49% identity), has UV/vis spectral features
at 323, 458, and 560 nm and the as-purified protein contains
1.2 irons per monomer as determined by ferrozine quantitation,
consistent with the enzyme having a [2Fe-2S] cluster (Figure
4).49,51 SpnQ also requires a reductase for activity. However,
no E3 gene homologue can be found in the spinosyn gene cluster.
SpnQ is therefore thought to catalyze a 3-dehydration reaction
in a mechanistically similar manner to that of E15,22 but employs
cellular reductases ferredoxin/ferredoxin reductase or flavodoxin/
flavodoxin reductase rather than a specific partner reductase,
as in the E1 case. Interestingly, in the absence of an electron
source and in the presence of L-glutamate, SpnQ catalyzes a
transamination reaction, converting 12 to TDP-4-amino-2,4,6-
trideoxy-D-glucose (23, retention time of 15.8 min in Figure 3,
trace e).31 The SpnQ homologue ColD (26% identity) lacks an
iron-sulfur cluster and does not require a reductase. Instead ColD
catalyzes the 3-dehydration using L-glutamate to regenerate
PMP.23 In the absence of E3 and in the presence of L-glutamate,
E1 has also been shown to carry out a 3-dehydration reaction
rather than a transamination reaction,49 making it akin to ColD
in this respect. Clearly, SpnQ, E1, and ColD each have unique
catalytic properties.
To our knowledge, the TDP-4-aminosugar 23 is neither a
characterized nor a predicted intermediate in any biosynthetic
pathway, and therefore its preparation using SpnQ warrants
optimization. It was found that L-glutamate is 20-fold more
efficient than L-aspartate as an amino donor. Further analysis
also showed that the reaction proceeds optimally at pH 7.5 and
at 37 °C. Next, the ability of SpnQ to process an alternate
substrate 10 was investigated by incubating 10 with SpnQ in
the presence of ferredoxin, ferredoxin reductase, and NADPH.
Unfortunately, no consumption of substrate or formation of any
new products was detected by HPLC even after prolonged
incubation, indicating that 10 is not a substrate for SpnQ. Finally,
the steady-state kinetics parameters for SpnQ were determined
using an assay similar to that previously reported for E1/E352 in
which NADPH consumption was monitored spectrophotometri-
cally in the presence of 12, ferredoxin, and ferredoxin reductase.
Figure 5. UV/visible spectra of (a) as-isolated SpnR (19.4 µM) and (b)
SpnR reconstituted with PLP (17.0 µM). See Experimental Section for
details on reconstitution.
A kcat value of 2.6 min-1 and a Km value of 49 µM for 12 were
obtained (Table 1). These values are similar to those determined
for E1 (kcat ) 3.5 min-1, Km of 44 µM)52 under similar
conditions. Based on the kinetic parameters of the ColD reaction
(kcat ) 36 min-1, Kmapp of 49 µM),23 the catalytic efficiency of
ColD is apparently higher than those of SpnQ and E1.
Purification and Characterization of SpnR. To assess the
catalytic activity of SpnR, the proposed TDP-4-keto-2,3,6-
trideoxy-D-glucose (13) 4-aminotransferase, spnR, was cloned
into the pET24b(+) expression vector giving a C-terminal His6-
tagged protein upon expression in E. coli. The recombinant
SpnR protein was purified to homogeneity after sequential Ni-
NTA and Sephacryl S-200 chromatographic separations. The
yield of SpnR was 80 mg of protein from 3 L of culture. A
molecular mass of 43 kDa of SpnR estimated by SDS-PAGE
(Figure 2, lane 4) correlates well with the predicted value of
43,356 Da calculated from the deduced amino acid sequence.
A molecular mass of 85.9 kDa, as judged by gel filtration
chromatography, indicates that the SpnR exists as a dimer in
solution. The UV-visible spectrum of the purified SpnR showed
only weak absorbance at 412 nm corresponding to less than
0.1 equiv of PLP cofactor per monomer (Figure 5, spectrum
a). After reconstitution with excess PLP, SpnR was determined
to have nearly 1 equiv of PLP per monomer (Figure 5, spectrum
b). These results indicated that exogenous PLP is required in
the assay for SpnR to be fully active.
Chemical Synthesis of the Phosphonate Mimic of SpnR
Product 24. As depicted in Scheme 4, preparation of this
compound started from (R)-2-benzoxypropionaldehyde (28). The
key intermediate 31 was prepared in three steps, involving the
diastereoselective chain extension to a (5R,6R)-diol (29),53 azide
displacement under modified Mitsunobu conditions to give 30,54
and asymmetric dihydroxylation of the terminal olefin to yield
31.55 After protective group manipulations, the linear azide 32
(52) Lei, Y.; Ploux, O.; Liu, H.-w. Biochemistry 1995, 34, 4643-4654.
(53) Roush, W. R.; Bennett, C. E.; Roberts, S. E. J. Org. Chem. 2001, 66, 6389-
6393.
(54) Lal, B.; Pramanik, B. N.; Manhas, M. S.; Bose, A. K. Tetrahedron Lett.
1977, 23, 1977-1980.
(55) Sharpless, K. B.; Amberg, W.; Bennani, Y. L.; Crispino, G. A.; Hartung,
J.; Jeong, K. S.; Kwong, H. L.; Morikawa, K.; Wang, Z. M.; Xu, D.; Zhang,
X. L. J. Org. Chem. 1992, 57, 2768-2771.
(49) Wu, Q.; Liu, Y. N.; Chen, H.; Molitor, E. J.; Liu, H.-w. Biochemistry 2007,
46, 3759-3767.
(50) Agnihotri, G.; Liu, Y.-n.; Paschal, B. M.; Liu, H.-w. Biochemistry 2004,
43, 14265-14274.
(51) Matsubara, H.; Katsube, Y.; Wada, K. Iron-Sulfur Protein Research; Japan
Scientific Societies Press: Tokyo, 1987.
9
4964 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

TDP-D-Forosamine Biosynthesis in the Spinosyn Pathway
A R T I C L E S
and L-glutamate. HPLC analysis indicated that the conversion
of 27 to 24 was almost complete within 4 h under the assay
conditions (Figure 6, trace b). The ability of SpnR to catalyze
transamination of its natural substrate 13 was also tested.
Compound 13 was first generated from 12 using SpnQ,
ferredoxin, ferredoxin reductase, and NADPH. Subsequent
incubation of the filtrate from the SpnQ reaction with SpnR,
PLP, and L-glutamate led to the depletion of 13 and to the
formation of a new peak (retention time of 9.2 min, Figure 3,
trace f), corresponding to the SpnR product TDP-4-amino-
2,3,4,6-tetradeoxy-D-glucose (14).
Purification and Characterization of SpnS. To establish
the function of SpnS as the N,N-dimethyltransferase catalyzing
the final step in TDP-D-forosamine (16) biosynthesis, we cloned
the spnS gene into the pET24b(+) expression vector producing
a C-terminal His6-tagged protein upon expression in E. coli.
SpnS was purified to homogeneity by sequential Ni-NTA and
Sephacryl S-200 chromatographic steps, yielding 50 mg of pure
protein from a 6 L culture. The absorption spectrum of the
purified protein exhibits no absorbance above 300 nm. SDS-
PAGE analysis of purified SpnS (Figure 2, lane 5) shows that
the protein has a molecular mass of 28 kDa, which correlates
well with the predicted molecular weight of 28 730 Da deduced
from the amino acid sequence. A native molecular mass of 62.1
kDa, as judged by gel filtration chromatography, indicates that
SpnS exists as a dimeric protein.
Figure 6. SpnR and SpnS reactions using phosphonate substrate mimics.
Samples a and b were analyzed using a Dionex Carbopac PA1 column.
Samples c and d were analyzed using the SAX Adsorbosphere column. (a)
Incubation of amino phosphonate mimic 24 (1 mM) with SpnR (10 µM),
R-ketoglutarate (30 mM), and PLP (40 mM) in 50 mM KH2PO4 buffer
(pH 7.5); (b) incubation of keto phosphonate mimic 27 (1 mM) with SpnR
(10 µM), L-glutamate (10 mM), and PLP (40 mM) in 50 mM KH2PO4
buffer (pH 7.5); (c) incubation of amino phosphonate mimic 24 (1 mM)
with SpnS (10 µM), SAM (2 mM), MgCl2 (2 mM), DTT (2 mM) in 50
mM KH2PO4 buffer (pH 7.5); (d) purified N-monomethylamino phosphonate
mimic 25 from preparative scale SpnS reaction using substrate 24.
was converted to the cyclic C-glycoside 33 by an intramolecular
SN2 substitution reaction.56 Less than 5% cyclized product with
the incorrect stereochemistry at the pseudoanomeric center was
obtained, and this was easily separated from 33 by silica gel
chromatography. The azide functionality was then converted
to the Boc-protected amino group under reductive amidation
conditions,57 and the C-P linkage was introduced under
Arbuzov conditions58 to give phosphonate 35. Trimethylsilyl
bromide was used to remove both the phosphonate ester and
the Boc protecting groups, and the resulting phosphonic acid
36 was coupled with thymidine 5′-monophosphomorpholidate
in the presence of 1H-tetrazole to give 24, which was purified
by size-exclusion chromatography on Bio-Gel P2 extra fine
resin.59
Catalytic Properties of SpnR. Incubation of the synthesized
phosphonate 24 with SpnR in the presence of PLP and
R-ketoglutarate led to the formation of the 4-keto product 27
(Figure 6, trace a), which was isolated and characterized by
NMR spectroscopy and high-resolution mass spectrometry.30
The phosphonate mimic was used because of the natural SpnR
substrate 13 is unstable. Interestingly, SpnR was also able to
catalyze transamination of 12, resulting in formation of TDP-
4-amino-2,4,6-trideoxy-D-glucose (23, Figure 3, trace g), the
same compound made by SpnQ in the presence of PLP and
L-glutamate. Several additional experiments were also performed
to characterize the PLP-dependence and cosubstrate specificity
of the SpnR-catalyzed reaction. First, replacement of the
R-ketoglutarate cosubstrate with pyruvate in the transamination
reaction (24 f 27) resulted in a 2-fold decrease in the rate of
product formation, indicating that pyruvate was less efficient
than R-ketoglutarate in the reaction, but could nevertheless
substitute for it. To test the aminotransferase activity of SpnR
in the forward direction, the ketosugar phosphonate mimic 27,
prepared using SpnR and 24, was incubated with SpnR, PLP,
Catalytic Properties of SpnS. Because of the difficulty of
preparing sufficient quantities of the SpnS substrate 14 for NMR
spectroscopic analysis of the expected product TDP-D-foro-
samine (16), SpnS was assayed using previously synthesized
phosphonate mimic 24. In our initial HPLC assays of SpnS with
24 and SAM, no new product peaks were detected. A survey
of various assay conditions led to the discovery that Mg2+ was
absolutely required for SpnS activity, and that 2 mM MgCl2
was optimal for catalytic turnover. Under these conditions, time-
dependent disappearance of substrate was observed, but the
product peaks were not clearly discernible because they
overlapped with those of SAM and SAH (retention times of
2.6-5.5 min). Analysis of the SpnS reaction mixture by SAX
ion exchange chromatography allowed separation of the product
peak (retention time of 20.2 min) from those of SAM and SAH
(Figure 6, trace c). In order to characterize this compound, a
preparative scale reaction using 24 was carried out (Figure 6,
trace d). The new compound was purified by Biogel P2
chromatography and analyzed by 1H, 13C, and 31P NMR
spectroscopy and high-resolution mass spectrometry. Interest-
ingly, the new compound was the N-monomethylated product
25 rather than the expected dimethylated product 26.
To test the dimethyltransferase activity of SpnS, 25 was
incubated with SpnS, SAM, and MgCl2, and the formation of
compound 26, which coeluted from the Dionex CarboPac
column with SAM (retention time of 2.6-5.5 min), was verified
by mass spectrometry. Our results indicated that SpnS is indeed
capable of N-methylation of 25 (Table 2). Subsequent time
course study of SpnS reaction using HPLC equipped with a
SAX column (Figure 7) revealed that using an initial concentra-
tion of 1 mM of 24, the concentration of 25 reached a steady
state after 1 h, while that of 24 continued to decrease for an
additional 3 h. This implied that 26 was formed in a time-
dependent manner from 25. Based on the observed rate of
(56) Solladie, G.; Arce, E.; Bauder, C.; Carreno, M. C. J. Org. Chem. 1998,
63, 2332-2337.
(57) Besse, P.; Veschambre, H.; Dickman, M.; Chenevert, R. J. Org. Chem.
1994, 59, 8288-8291.
(58) Arbuzov, B. A. Pure Appl. Chem. 1964, 9, 307-335.
(59) Zhao, Z.; Liu, H.-w. J. Org. Chem. 2001, 66, 6810-6815.
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4965

A R T I C L E S
Hong et al.
Table 2. Summary of Kinetic Parameters of First and Second
Methylation Reactions of Characterized TDP-sugar
N,N-Dimethyltransferases SpnS, DesVI, and TylM1
useful tailoring reaction which could be applied to modify a
variety of NDP-4-aminosugars.
The spinosyn pathway forosaminyltransferase SpnP has
previously been shown via biotransformation experiments using
an engineered S. erythraea host expressing spnP, to catalyze
the attachment of the sugars L-mycarose and D-glucose to the
spinosyn aglycone in ViVo.60 Interestingly, several naturally
occurring spinosyn congeners have forosamine derivatives,
including N-monodesmethyl-D-forosamine (spinosyns B, M, N,
and R), 4-amino-2,3,4,6-tetradeoxy-D-glucose (spinosyn C), and
its C-4 epimer (spinosyn G) at C-17,15 suggesting that SpnP is
also responsible for the transfer of these sugars to the spinosyn
aglycone. Our finding that the N-monodesmethylated TDP-D-
forosamine mimic 25 accumulates in significant quantities in
the SpnS in Vitro reaction suggests that TDP-N-monodesmethyl-
D-forosamine (15, Scheme 2) may also be present in significant
quantities in ViVo. Accumulation of 15 combined with the
inherent substrate flexibility of SpnP could together be respon-
sible for the formation of spinosyns having N-monodesmethyl-
D-forosamine. A more systematic investigation of the substrate
flexibility of SpnP toward various TDP-sugars has not been
carried out but is certainly warranted given the potential of using
this enzyme for combinatorial biosynthesis.
The mechanism of NDP-hexose C-3 deoxygenation has been
the focus of intense study over the past 30 years.5,22 The PMP-
dependent 3-dehydrase E1 from Y. pseudotuberculosis, which
acts on a 4-keto-6-deoxyhexose substrate, was the first char-
acterized enzyme having this activity.61 E1 shares modest
sequence similarity with PLP-dependent aminotransferases,
indicating an evolutionary link between the two classes of
enzymes.49 However, the Schiff base-forming lysine residue
conserved in vitamin B6-dependent aminotransferases is replaced
by a histidine in E1 and its homologues. Mechanistic studies of
E1 have elucidated the details of its catalytic cycle and have
made it the prototype for understanding how nature performs
hexose C-3 deoxygenation. Functional elucidation and initial
biochemical characterization of SpnQ illustrate that nature also
employs an “E1-like” strategy for the C-3 deoxygenation of 2,6-
dideoxyhexoses, although SpnQ distinguishes itself from E1 by
using general cellular reductases rather than a conserved
reductase partner (E3) in its catalysis.
In the E1 reaction, after initial Schiff base formation between
37 and PMP, C-4′ of the PMP-ketimine complex 38 is
deprotonated, triggering the expulsion of the C-3 hydroxyl group
to form the PMP-∆3,4-glycoseen intermediate 39 (Scheme 5).50
This intermediate undergoes two-electron reduction to form the
Schiff base intermediate 40, which is hydrolyzed to give product
41 and regenerate PMP (Scheme 5, Path A). Recent mechanis-
tic23 and structural24 studies of the E1 homologue ColD highlight
an interesting variation on the E1 mechanistic theme. ColD lacks
electron-transfer capability but carries out C-3 deoxygenation
by directly hydrolyzing the PMP-∆3,4-glycoseen intermediate
39 to give enamine 42 and PLP. Enzyme-mediated tautomer-
ization results in imine 43, which is hydrolyzed to release the
3-deoxygenated product 41 and ammonia (Scheme 5, Path B).
PMP is regenerated from PLP by a transamination reaction
catalyzed by ColD using L-glutamate. SpnQ is thought to be
first methylation
second methylation
k
cat/Km
k
cat/Km
1
1
1
1
enzyme
k
cat (min-
)
Km
(
µ
M) (min-1 µM-
)
k
cat (min-
)
Km
(
µ
M) (min-1 µM-
)
SpnS
DesVI
TylM1
1.6a
92.0
9.1
ND
307.4
59.4
ND
0.299
0.167
1.6 a
ND
32.5
ND
ND
46.8
ND
ND
0.694
a The kcat values for the SpnS-catalyzed reaction are kobs values.
Figure 7. Time course of the SpnS (10 µM)-catalyzed reaction using 24
(1 mM) as the substrate in the presence of SAM (2 mM), MgCl2 (2 mM),
and DTT (2 mM) in 50 mM KH2PO4 (pH 7.5). Integrations of the HPLC
peaks of substrate and monomethylated product 25, and the inferred
concentration of the dimethylated product 26 are plotted versus time: (O)
24, (b) 25, and (0) 26 (see Experimental Section for details).
substrate depletion and on the inferred rate of dimethylated
product formation, kobs values of 1.6 min-1 were calculated for
both the first and second methylation reactions. These experi-
ments provide compelling evidence that SpnS is the TDP-4-
amino-2,3,4,6-tetradeoxy-D-glucose 4-N,N-methyltransferase which
converts 14 to TDP-D-forosamine (16), and that N-methylation
occurs in a stepwise manner involving release of monomethy-
lated product from the active site, as has been observed in the
characterization of the SpnS homologue TylM1.34
Discussion
The sugar D-forosamine is interesting because of its highly
deoxygenated nature and its presence in several biologically
important natural products, including the commercially available,
environmentally benign insecticide spinosyns. Through studies
of the biosynthesis of TDP-D-forosamine (16) in the spinosyn
pathway of Saccharopolyspora spinosa, we have elucidated
nature’s strategy for the synthesis of this highly deoxygenated
sugar. This work also provides the foundation for future
exploration of the biosynthesis of other 2,3,6-trideoxy-and
2,3,4,6-tetradeoxyhexoses, which are found in many bioactive
natural products.
The functional elucidation of SpnO and SpnN, and the use
of SpnN to synthesize the key TDP-deoxysugar intermediate
12 are important milestones in efforts to enzymatically synthe-
size diverse NDP-deoxysugar donors for in Vitro construction
of new glycoforms of secondary metabolites. The results of
substrate specificity studies of SpnQ and SpnR have illustrated
the biosynthetic capabilities as well as the limitations of these
enzymes and provided a method for preparing the non-natural
TDP-sugar 23, which may be useful in glycodiversification
efforts. The in Vitro demonstration of the 4-N-mono- and 4-N,N-
dimethyltransferase activities of SpnS provided a synthetically
(60) Gaisser, S.; Martin, C. J.; Wilkinson, B.; Sheridan, R. M.; Lill, R. E.;
Weston, A. J.; Ready, S. J.; Waldron, C.; Crouse, G. D.; Leadlay, P. F.;
Staunton, J. Chem. Commun. 2002, 618-619.
(61) Rubenstein, P. A.; Strominger, J. L. J. Biol. Chem. 1974, 249, 3776-3781.
9
4966 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

TDP-D-Forosamine Biosynthesis in the Spinosyn Pathway
A R T I C L E S
Table 3. Native Molecular Weights and Aggregation States of Spinosyn Pathway Enzymes
enzyme
SpnO
SpnN
SpnQ
SpnR
SpnS
monomer molecular weight (KDa)
native molecular weight (KDa)
native aggregation state
55
96
dimer
38
38
monomer
51
109
dimer
43
86
dimer
28
62
dimer
Scheme 5. Proposed Mechanistic Routes for PMP-Dependent 3-Dehydrases under Various Conditionsa
a Path A represents the “reduction” route which occurs in the E1 and SpnQ reactions under normal conditions. Path B represents the “hydrolysis” route
undergone by ColD under normal conditions and by E1 under “ColD-like” conditions. Path C represents the “transaminase-like” route likely undergone by
SpnQ under “ColD-like” conditions.
mechanistically similar to E1, with catalysis proceeding via Path
A in the presence of an electron source. However, in the absence
of an electron source and in the presence of L-glutamate, SpnQ
surprisingly catalyzes transamination via Path C, producing the
3-hydroxy-4-aminosugar 45. Under the same conditions, E1
catalyzes a 3-dehydration reaction to give 41, presumably via
Path B.31 Thus, although E1, ColD, and SpnQ are homologous
NDP-hexose 3-dehydrases, each enzyme has distinct biochemi-
cal properties which clearly warrant further mechanistic and
structural study.
Recently, two structures of ColD from E. coli were solved.24
This enzyme, like E1, has a conserved active site His residue
(His188). Interestingly, other than His188, the ColD active site
is devoid of catalytic residues that are properly positioned for
the various protonation and deprotonation steps required to
complete the catalytic cycle. His188 is therefore proposed to
catalyze essentially all acid-base chemistry during ColD
catalysis: deprotonation of C-4′ of 38, protonation of the leaving
C-3 hydroxyl to generate 39, activation of the hydrolytic water
to form 42, and stereospecific protonation of C-3 of the enamine
to form 43.24 Although structural information on E1 and SpnQ
is not yet available, it is tempting to speculate that the divergent
behaviors of E1 and SpnQ under “ColD-like” conditions are
based on subtle differences in the positioning of the catalytic
residue(s) in the active sites of the two proteins. In the absence
of a reductase, 38 and 39 would be in equilibrium in the active
site of E1. The catalytic residue(s) may be appropriately
positioned to activate a water molecule for hydrolysis of 39
and to reprotonate at C-3 of 42 (Path B), resulting in 3-dehydra-
tion. In SpnQ, the catalytic residue(s) may be in the correct
orientation to facilitate isomerization between 38 and 44, leading
to the formation of 45 after hydrolysis (Path C). Recent success
in efforts to obtain a crystal structure of E135 may provide further
insights. With sufficient structural information in hand, it may
be possible to design 3-dehydrase mutants in which aminotrans-
ferase, reductase-dependent 3-dehydrase, and reductase-inde-
pendent 3-dehydrase activities are rationally interconverted.
Acknowledgment. This work was supported in part by the
National Institutes of Health Grants (GM35906, 54346).
JA0771383
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4967